A contamination model for approximate stochastic order

Size: px
Start display at page:

Download "A contamination model for approximate stochastic order"

Transcription

1 A contamination model for approximate stochastic order Eustasio del Barrio Universidad de Valladolid. IMUVA. 3rd Workshop on Analysis, Geometry and Probability - Universität Ulm 28th September - 2dn October 2015, Ulm Eustasio del Barrio Testing approximate stochastic order 1 / 37

2 Outline Outline 1 Stochastic order, testability and relaxed versions of s. o. 2 Inference for approximate stochastic order 3 Implementation, simulation & data example Eustasio del Barrio Testing approximate stochastic order 2 / 37

3 Stochastic order, testability and relaxed versions of s. o. The stochastic order model π(fn,gm) = π(gm,fn) = EDF's Gm Age 10 Boys Fn Age 10 Girls heights Data: National Health and Nutrition Examination Survey Empirical d.f. s for boys and girls at age 10. Are girls taller than boys? Stochastic order (Lehmann, 1955): P, Q probs. on R with d.f. s F, G For NHANES data, P 10 st Q 10? P st Q if F (x) G(x), x R Eustasio del Barrio Testing approximate stochastic order 3 / 37

4 Stochastic order, testability and relaxed versions of s. o. Testing stochastic order Common testing problems in literature (P st Q F st G) a) H 0 : F = G vs H a : F < st G b) H 0 : F st G vs H a : F st G c) H 0 : F st G vs H a : F st G Problem a) focus on statistical evidence for strict relation assumes stochastic order holds both H 0 and H a can be false (here focus on b), c)) Eustasio del Barrio Testing approximate stochastic order 4 / 37

5 Stochastic order, testability and relaxed versions of s. o. Testing stochastic order Common testing problems in literature (P st Q F st G) Problem a) a) H 0 : F = G vs H a : F < st G b) H 0 : F st G vs H a : F st G c) H 0 : F st G vs H a : F st G focus on statistical evidence for strict relation assumes stochastic order holds both H 0 and H a can be false (here focus on b), c)) Problem b) testing for stochastic dominance (McFadden, 1989; Mosler, 1995; Anderson, 1996, Davidson & Duclos, 2000; Linton et al., 2005, 2010,... ) goodness-of-fit problem absence of evidence against s. o. as minimal requirement for a) lack of evidence against H 0 not evidence for F st G Eustasio del Barrio Testing approximate stochastic order 4 / 37

6 Stochastic order, testability and relaxed versions of s. o. Testing stochastic order Problem c) H 0 : F st G vs H a : F < st G: assessing stochastic order rejection provides convincing evidence of F < st G Eustasio del Barrio Testing approximate stochastic order 5 / 37

7 Stochastic order, testability and relaxed versions of s. o. Testing stochastic order Problem c) H 0 : F st G vs H a : F < st G: assessing stochastic order rejection provides convincing evidence of F < st G Unfortunately, no good test for b) exists: Assume X 1,..., X n i.i.d. F < st G Φ an α-level test (E H Φ(X 1,..., X n ) α, H H 0 ) Take x m s.t. G(x m ) > 1 1 m, H m s.t. H m (x m ) = 0 Set F m = (1 1 m )F + 1 m H m; F m st G α E Fm Φ(X 1,..., X n ) (1 1 m )n E F Φ(X 1,..., X n ) Take m no data test (reject H 0 with prob α regardless data) is UMP! Berger, 1988 (one-sample); Davidson & Duclos, 2013 (two-sample) Eustasio del Barrio Testing approximate stochastic order 5 / 37

8 Stochastic order, testability and relaxed versions of s. o. Uniformly consistent tests Uniformly consistent tests X 1, X 2,... i.i.d. P with values in X A 0,n (A 1,n ) acceptance (critical) set for H n against K n based on X 1,..., X n Test uniformly (exponentially) consistent if for some r, r > 0 sup P n (A 1,n ) e rn, P H n sup P n (A 0,n ) e r n P K n Consider the testing problem H : P = P 0 vs K : d(p, P 0 ) > δ If d dominates d T V (Barron, 1989) and P 0 not discrete, no uniformly consistent test of H vs K Eustasio del Barrio Testing approximate stochastic order 6 / 37

9 Stochastic order, testability and relaxed versions of s. o. Uniformly consistent tests Here we propose A relaxed version of stochastic order for which we can expect to get statistical evidence A consistent procedure for gathering that evidence Which is exponentially uniformly consistent (with due corrections) Some of our relaxations does hold Deviation from stochastic order measured through required level of relaxation Easy interpretation Eustasio del Barrio Testing approximate stochastic order 7 / 37

10 Stochastic order, testability and relaxed versions of s. o. Relaxations of stochastic order A relaxation of stochastic order (Arcones et al., 2002) θ(p, Q) := P[X Y ] X, Y independent r.v. s with laws P, Q, resp. P sp Q (stochastically precedes) if θ(p, Q) 1 2 Stochastic ordering implies stochastic precedence: if P st Q P(X Y ) = (1 G(x ))df (x) (1 F (x ))df (x) = P(X X ) 1 2, X independent copy of X Eustasio del Barrio Testing approximate stochastic order 8 / 37

11 Stochastic order, testability and relaxed versions of s. o. Relaxations of stochastic order A relaxation of stochastic order (Arcones et al., 2002) θ(p, Q) := P[X Y ] X, Y independent r.v. s with laws P, Q, resp. P sp Q (stochastically precedes) if θ(p, Q) 1 2 Stochastic ordering implies stochastic precedence: if P st Q P(X Y ) = (1 G(x ))df (x) (1 F (x ))df (x) = P(X X ) 1 2, X independent copy of X Stochastic precedence a less restrictive assumption Eustasio del Barrio Testing approximate stochastic order 8 / 37

12 Stochastic order, testability and relaxed versions of s. o. Relaxations of stochastic order A relaxation of stochastic order (Arcones et al., 2002) θ(p, Q) := P[X Y ] X, Y independent r.v. s with laws P, Q, resp. P sp Q (stochastically precedes) if θ(p, Q) 1 2 Stochastic ordering implies stochastic precedence: if P st Q P(X Y ) = (1 G(x ))df (x) (1 F (x ))df (x) = P(X X ) 1 2, X independent copy of X Stochastic precedence a less restrictive assumption But P sp Q equivalent to median(y X) 0, roughly change in location Eustasio del Barrio Testing approximate stochastic order 8 / 37

13 Stochastic order, testability and relaxed versions of s. o. Relaxations of stochastic order A relaxation of stochastic order (Arcones et al., 2002) θ(p, Q) := P[X Y ] X, Y independent r.v. s with laws P, Q, resp. P sp Q (stochastically precedes) if θ(p, Q) 1 2 Stochastic ordering implies stochastic precedence: if P st Q P(X Y ) = (1 G(x ))df (x) (1 F (x ))df (x) = P(X X ) 1 2, X independent copy of X Stochastic precedence a less restrictive assumption But P sp Q equivalent to median(y X) 0, roughly change in location (different, but similar nature as E(Y X) 0) Eustasio del Barrio Testing approximate stochastic order 8 / 37

14 Stochastic order, testability and relaxed versions of s. o. Tolerance zones around false models False model assessment Assume model F is false (X P, P / F) Is model F an adequate approximation for the data? for P? P θ F is an adequate approximation for the data, X 1,..., X n, if a typical sample of size n from P θ looks like the data Data features (Davies, 1995) Credibility indices (Lindsay & Liu, 2009) P θ F gives an adequate description of P if d(p, P θ ) τ d = χ 2 distance (Hodges & Lehmann, 1954) d = Euclidean distance (Dette & Munk, 2003) d = smallest π such that P = (1 π)p θ + πr (Rudas et al. (1994); Ae-dB-C-M, 2008, 2010, 2011, 2012; Liu & Lindsay, 2009; Cerioli et al., 2012) Choice of τ a hard issue Interpretation of τ simpler for the π-index Eustasio del Barrio Testing approximate stochastic order 9 / 37

15 Essential model validation Stochastic order, testability and relaxed versions of s. o. Essential model validation Observe data X P, test H 1 P = (1 α)r + α P for some R F Observe ind. samples X P, Y Q, test H 2 P = (1 α)r + α P Q = (1 α)s + α Q for some (R, S) F Related problem of interest: Find α 0 = minimal α s.t. null model holds Eustasio del Barrio Testing approximate stochastic order 10 / 37

16 Essential model validation Stochastic order, testability and relaxed versions of s. o. Essential model validation Observe data X P, test H 1 P = (1 α)r + α P for some R F Observe ind. samples X P, Y Q, test H 2 P = (1 α)r + α P Q = (1 α)s + α Q for some (R, S) F Related problem of interest: Find α 0 = minimal α s.t. null model holds Example: the similarity model (AE-dB-C-M, 2012) P and Q α-similar, α [0, 1) if prob, R, s.t. { P = (1 α)r + α P Q = (1 α)r + α Q Eustasio del Barrio Testing approximate stochastic order 10 / 37

17 Essential model validation Stochastic order, testability and relaxed versions of s. o. Essential model validation Observe data X P, test H 1 P = (1 α)r + α P for some R F Observe ind. samples X P, Y Q, test H 2 P = (1 α)r + α P Q = (1 α)s + α Q for some (R, S) F Related problem of interest: Find α 0 = minimal α s.t. null model holds Example: the similarity model (AE-dB-C-M, 2012) P and Q α-similar, α [0, 1) if prob, R, s.t. { P = (1 α)r + α P Q = (1 α)r + α Q P, Q α-similar, d T V (P, Q) α (d T V (P, Q) = sup A P (A) Q(A) ) Eustasio del Barrio Testing approximate stochastic order 10 / 37

18 Essential model validation Stochastic order, testability and relaxed versions of s. o. Essential model validation Observe data X P, test H 1 P = (1 α)r + α P for some R F Observe ind. samples X P, Y Q, test H 2 P = (1 α)r + α P Q = (1 α)s + α Q for some (R, S) F Related problem of interest: Find α 0 = minimal α s.t. null model holds Example: the similarity model (AE-dB-C-M, 2012) P and Q α-similar, α [0, 1) if prob, R, s.t. { P = (1 α)r + α P Q = (1 α)r + α Q P, Q α-similar, d T V (P, Q) α (d T V (P, Q) = sup A P (A) Q(A) ) Here F = {(R, R)} and α 0,sim (P, Q) = d T V (P, Q) Eustasio del Barrio Testing approximate stochastic order 10 / 37

19 Stochastic order, testability and relaxed versions of s. o. Essential model validation Approximate stochastic order: P st,α Q if P = (1 α)r + α P Q = (1 α)s + α Q for some R st S Eustasio del Barrio Testing approximate stochastic order 11 / 37

20 Stochastic order, testability and relaxed versions of s. o. Essential model validation Approximate stochastic order: P st,α Q if P = (1 α)r + α P Q = (1 α)s + α Q for some R st S (F = {(R, S) : R st S}) Eustasio del Barrio Testing approximate stochastic order 11 / 37

21 Stochastic order, testability and relaxed versions of s. o. Essential model validation Approximate stochastic order: P st,α Q if P = (1 α)r + α P Q = (1 α)s + α Q for some R st S (F = {(R, S) : R st S}) (maybe s. o. too restrictive, but core of distribution fits model) Eustasio del Barrio Testing approximate stochastic order 11 / 37

22 Stochastic order, testability and relaxed versions of s. o. Essential model validation Approximate stochastic order: P st,α Q if P = (1 α)r + α P Q = (1 α)s + α Q for some R st S (F = {(R, S) : R st S}) (maybe s. o. too restrictive, but core of distribution fits model) Interest on minimal contamination level s.t. stochastic order model holds α 0 (P, Q) := inf{α : P st,α Q} Eustasio del Barrio Testing approximate stochastic order 11 / 37

23 Stochastic order, testability and relaxed versions of s. o. Trimming methods in essential model validation Trimmed Distributions (X, β) measurable space; P(X, β) prob. measures on (X, β), P P(X, β) { dr R α (P ) = R P(X, β) : R P, dp 1 } 1 α P -a.s. Proposition (a) R α (P ) is a convex set; α 1 α 2 R α1 (P ) R α2 (P ) (b) If α < 1 and (X, β) complete separable metric space then R α (P ) compact for weak convergence. (c) R R α (P ) iff P = (1 α)r + α P Eustasio del Barrio Testing approximate stochastic order 12 / 37

24 Stochastic order, testability and relaxed versions of s. o. Essential model validation & trimming Null models in essential model validation expressable in terms of trimmings Observe X P, test H 1 P = (1 α)r + α P for some R F H 1 holds iff R α (P ) F Observe indep. X P, Y Q test H 2 P = (1 α)r + α P Q = (1 α)s + α Q for some (R, S) F H 2 holds iff (R α (P ) R α (Q)) F If R(P ), F closed for metric d H 1 holds iff d(r α (P ), F) = 0 H 2 holds iff d(r α (P ) R α (Q), F) = 0 Eustasio del Barrio Testing approximate stochastic order 13 / 37

25 Stochastic order, testability and relaxed versions of s. o. Approximate stochastic order & trimming Trimmings mix well with stochastic order: For any P P α, P α in R π (P ) s. t. P α st R st P α for every R R α (P ) Eustasio del Barrio Testing approximate stochastic order 14 / 37

26 Stochastic order, testability and relaxed versions of s. o. Approximate stochastic order & trimming Trimmings mix well with stochastic order: For any P P α, P α in R π (P ) s. t. P α st R st P α for every R R α (P ) P α, P α easily computable Eustasio del Barrio Testing approximate stochastic order 14 / 37

27 Stochastic order, testability and relaxed versions of s. o. Approximate stochastic order & trimming Trimmings mix well with stochastic order: For any P P α, P α in R π (P ) s. t. P α st R st P α for every R R α (P ) P α, P α easily computable Recall P st,α Q iff P R α (P ), Q Rα (Q), s.t. P st Q Eustasio del Barrio Testing approximate stochastic order 14 / 37

28 Stochastic order, testability and relaxed versions of s. o. Approximate stochastic order & trimming Trimmings mix well with stochastic order: For any P P α, P α in R π (P ) s. t. P α st R st P α for every R R α (P ) P α, P α easily computable Recall P st,α Q iff P R α (P ), Q Rα (Q), s.t. P st Hence P st,α Q iff P α st Q α Q Eustasio del Barrio Testing approximate stochastic order 14 / 37

29 Stochastic order, testability and relaxed versions of s. o. Approximate stochastic order & trimming Trimmings mix well with stochastic order: For any P P α, P α in R π (P ) s. t. P α st R st P α for every R R α (P ) P α, P α easily computable Recall P st,α Q iff P R α (P ), Q Rα (Q), s.t. P st Hence P st,α Q iff P α st Q α Conclude from this α 0 (P, Q) = sup(g(x) F (x)) x Q Eustasio del Barrio Testing approximate stochastic order 14 / 37

30 Stochastic order, testability and relaxed versions of s. o. Approximate stochastic order & trimming Trimmings mix well with stochastic order: For any P P α, P α in R π (P ) s. t. P α st R st P α for every R R α (P ) P α, P α easily computable Recall P st,α Q iff P R α (P ), Q Rα (Q), s.t. P st Hence P st,α Q iff P α st Q α Conclude from this α 0 (P, Q) = sup(g(x) F (x)) x Q Equivalently, P st,α Q α 0 (P, Q) α Eustasio del Barrio Testing approximate stochastic order 14 / 37

31 Stochastic order, testability and relaxed versions of s. o. Approximate stochastic order & trimming Example 1. P = N(µ, σ), Q = N(ν, σ), µ > ν Now Q st P and, α 0 (P, Q) = (sup(g(x) F (x)) = 2Φ ( ) µ ν 2σ 1. x µ ν = 0.1σ P st,0.04 Q µ ν = 0.25σ P st, Q µ ν = 0.5σ P st, Q µ ν = σ P st, Q Eustasio del Barrio Testing approximate stochastic order 15 / 37

32 Stochastic order, testability and relaxed versions of s. o. Approximate stochastic order & trimming Example 1. P = N(µ, σ), Q = N(ν, σ), µ > ν Now Q st P and, α 0 (P, Q) = (sup(g(x) F (x)) = 2Φ ( ) µ ν 2σ 1. x µ ν = 0.1σ P st,0.04 Q µ ν = 0.25σ P st, Q µ ν = 0.5σ P st, Q µ ν = σ P st, Q Example 2. P = N(µ, σ), Q = N(ν, τ) Here α 0 (P, Q) depends on µ, ν, σ, τ Eustasio del Barrio Testing approximate stochastic order 15 / 37

33 Stochastic order, testability and relaxed versions of s. o. Approximate stochastic order & trimming Example 1. P = N(µ, σ), Q = N(ν, σ), µ > ν Now Q st P and, α 0 (P, Q) = (sup(g(x) F (x)) = 2Φ ( ) µ ν 2σ 1. x µ ν = 0.1σ P st,0.04 Q µ ν = 0.25σ P st, Q µ ν = 0.5σ P st, Q µ ν = σ P st, Q Example 2. P = N(µ, σ), Q = N(ν, τ) Here α 0 (P, Q) depends on µ, ν, σ, τ θ(p, Q) = µ ν (Arcones et al.,2002) For µ = ν, P sp Q regardless σ, τ Eustasio del Barrio Testing approximate stochastic order 15 / 37

34 Stochastic order, testability and relaxed versions of s. o. Approximate stochastic order & trimming Example 1. P = N(µ, σ), Q = N(ν, σ), µ > ν Now Q st P and, α 0 (P, Q) = (sup(g(x) F (x)) = 2Φ ( ) µ ν 2σ 1. x µ ν = 0.1σ P st,0.04 Q µ ν = 0.25σ P st, Q µ ν = 0.5σ P st, Q µ ν = σ P st, Q Example 2. P = N(µ, σ), Q = N(ν, τ) Here α 0 (P, Q) depends on µ, ν, σ, τ θ(p, Q) = µ ν (Arcones et al.,2002) For µ = ν, P sp Q regardless σ, τ But N(0, σ) takes values greater than N(0, 0) 50% of times! Eustasio del Barrio Testing approximate stochastic order 15 / 37

35 Inference for approximate stochastic order Estimation & testing Inference in approximate stochastic order models Assume X 1,..., X n i.i.d. P ; Y 1,..., Y m i.i.d. Q, independent samples Eustasio del Barrio Testing approximate stochastic order 16 / 37

36 Inference for approximate stochastic order Estimation & testing Inference in approximate stochastic order models Assume X 1,..., X n i.i.d. P ; Y 1,..., Y m i.i.d. Q, independent samples Goals (a) For a fixed α, test H 0 : P st,α Q vs. H 0 : P st,α Q (b) For a fixed α, test H 0 : P st,α Q vs. H 0 : P st,α Q (c) Estimation/confidence intervals/confidence bounds for α 0 (P, Q) Eustasio del Barrio Testing approximate stochastic order 16 / 37

37 Inference for approximate stochastic order Estimation & testing Inference in approximate stochastic order models Assume X 1,..., X n i.i.d. P ; Y 1,..., Y m i.i.d. Q, independent samples Goals (a) For a fixed α, test H 0 : P st,α Q vs. H 0 : P st,α Q (b) For a fixed α, test H 0 : P st,α Q vs. H 0 : P st,α Q (c) Estimation/confidence intervals/confidence bounds for α 0 (P, Q) Recall P st,α Q α 0 (P, Q) α; reformulate (a), (b) as (a) H 0 : α 0 (P, Q) α vs. H a : α 0 (P, Q) > α (testing against approximate s.o.) (b ) H 0 : α 0 (P, Q) α vs. H a : α 0 (P, Q) < α (testing for approximate s.o.) Eustasio del Barrio Testing approximate stochastic order 16 / 37

38 Inference for approximate stochastic order Asymptotic theory Assume F and G continuous; n = m F n, G n empirical d.f. s Theorem α 0 (F n, G n ) a.s. α 0 (F, G), n(α0 (F n, G n ) α 0 (F, G)) w B(F, G) with B(F, G) = B 1, B 2 independent Brownian Bridges; sup (B 1 (F (x)) B 2 (G(x))), x Γ(F,G) Γ(F, G) := {x R : F (x) G(x) = α 0 (F, G)} A bootstrap version also available, but slow approximation (support estimation) Eustasio del Barrio Testing approximate stochastic order 17 / 37

39 Inference for approximate stochastic order Asymptotic theory Quantiles of B(F, G) depend on F, G in a complex way Define B α = B(U(α, 1 + α), U(0, 1)), 0 α 1 Eustasio del Barrio Testing approximate stochastic order 18 / 37

40 Inference for approximate stochastic order Asymptotic theory Quantiles of B(F, G) depend on F, G in a complex way Define B α = B(U(α, 1 + α), U(0, 1)), 0 α 1 B α = sup (B 1 (t) B 2 (t α)) α t 1 P ( B 0 > 2t) = e t2 /2, α = 0, 0.1,..., 0.5 Eustasio del Barrio Testing approximate stochastic order 18 / 37

41 Inference for approximate stochastic order Asymptotic theory Bounds for asymptotic quantiles K β (F, G) (resp. K β (α)) β-quantile of B(F, G) (resp. Bα ) K β (F, G) K β (α(f, G)), β (0, 1) If β (0, 1 2 ] σ(f, G, α(f, G))Φ 1 (β) K β (F, G), where σ(f, G, α(f, G)) = min t T (F,G,α(F,G)) σ t, T (F, G, α(f, G)) = {t : x s.t. F (x) = t, G(x) = t α(f, G)} and σ 2 t = t(1 t) + (t α(f, G))(1 t + α(f, G)) Eustasio del Barrio Testing approximate stochastic order 19 / 37

42 Inference for approximate stochastic order Testing against essential stochastic order Testing against essential stochastic order Consider H 0 : α 0 (F, G) α, vs. H a : α 0 (F, G) > α Eustasio del Barrio Testing approximate stochastic order 20 / 37

43 Inference for approximate stochastic order Testing against essential stochastic order Testing against essential stochastic order Consider H 0 : α 0 (F, G) α, vs. H a : α 0 (F, G) > α (equivalently, H 0 : F st,α G vs. H a : F st,α G) Eustasio del Barrio Testing approximate stochastic order 20 / 37

44 Inference for approximate stochastic order Testing against essential stochastic order Testing against essential stochastic order Consider H 0 : α 0 (F, G) α, vs. H a : α 0 (F, G) > α (equivalently, H 0 : F st,α G vs. H a : F st,α G) Reject H 0 if n(α0 (F n, G n ) α) > K 1 β (α), K 1 β (α) = 1 β quantile of B(α) Eustasio del Barrio Testing approximate stochastic order 20 / 37

45 Inference for approximate stochastic order Testing against essential stochastic order Testing against essential stochastic order Consider H 0 : α 0 (F, G) α, vs. H a : α 0 (F, G) > α (equivalently, H 0 : F st,α G vs. H a : F st,α G) Reject H 0 if n(α0 (F n, G n ) α) > K 1 β (α), K 1 β (α) = 1 β quantile of B(α) Theorem lim n sup P F,G ( n(α 0 (F n, G n ) α) > K 1 β (α)) (F,G) H 0 F 0 U(α, 1 + α), G 0 U(0, 1) = lim n P F 0,G 0 ( n(α 0 (F n, G n ) α) > K 1 β (α)) = β, Eustasio del Barrio Testing approximate stochastic order 20 / 37

46 Inference for approximate stochastic order Testing against essential stochastic order Nonasymptotic bounds Theorem If α 0 (F, G) < α and K 1 β (α) 0 then P F,G ( n(α 0 (F n, G n ) α) > K 1 β (α)) 2e n(α α0(f,g))2. If α 0 (F, G) > α then P F,G ( n(α 0 (F n, G n ) α) K 1 β (α)) e 2( n(α α 0(F,G)) K 1 β (α)) 2. Eustasio del Barrio Testing approximate stochastic order 21 / 37

47 Inference for approximate stochastic order Testing against essential stochastic order Nonasymptotic bounds Theorem If α 0 (F, G) < α and K 1 β (α) 0 then P F,G ( n(α 0 (F n, G n ) α) > K 1 β (α)) 2e n(α α0(f,g))2. If α 0 (F, G) > α then P F,G ( n(α 0 (F n, G n ) α) K 1 β (α)) e 2( n(α α 0(F,G)) K 1 β (α)) 2. Test is u.e.c. for H 0 : α 0 (F, G) α vs. H a : α 0 (F, G) > α if α < α < α Compute sample sizes to guarantee given power against fixed alternatives Eustasio del Barrio Testing approximate stochastic order 21 / 37

48 Inference for approximate stochastic order Testing for essential stochastic order Testing for essential stochastic order Consider H 0 : α 0 (F, G) α, vs. H a : α 0 (F, G) < α Eustasio del Barrio Testing approximate stochastic order 22 / 37

49 Inference for approximate stochastic order Testing for essential stochastic order Testing for essential stochastic order Consider H 0 : α 0 (F, G) α, vs. H a : α 0 (F, G) < α (equivalently, H 0 : F st,α G if α < α vs. H a : F st,α G for some α < α) Eustasio del Barrio Testing approximate stochastic order 22 / 37

50 Inference for approximate stochastic order Testing for essential stochastic order Testing for essential stochastic order Consider H 0 : α 0 (F, G) α, vs. H a : α 0 (F, G) < α (equivalently, H 0 : F st,α G if α < α vs. H a : F st,α G for some α < α) Reject H 0 if n(α0 (F n, G n ) α) < σ α Φ 1 (β), where σ 2 α = 1 α2 2, (assume β < 1 2 ) Eustasio del Barrio Testing approximate stochastic order 22 / 37

51 Inference for approximate stochastic order Testing for essential stochastic order Testing for essential stochastic order Consider H 0 : α 0 (F, G) α, vs. H a : α 0 (F, G) < α (equivalently, H 0 : F st,α G if α < α vs. H a : F st,α G for some α < α) Reject H 0 if n(α0 (F n, G n ) α) < σ α Φ 1 (β), where σ 2 α = 1 α2 2, (assume β < 1 2 ) Theorem lim n F 0 1 α 1+α 2 U(0, sup P F,G ( n(α 0 (F n, G n ) α) < σ α Φ 1 (β)) (F,G) H 0 = lim n P F 0,G 0 ( n(α 0 (F n, G n ) α) < σ α Φ 1 (β)) = β, 2 ) + 1+α 2 U( 1+α 2, 1 + α(1 α) 2 ), G 0 U(0, 1) Eustasio del Barrio Testing approximate stochastic order 22 / 37

52 Inference for approximate stochastic order Testing for essential stochastic order Nonasymptotic bounds Theorem If α 0 (F, G) > α then P F,G ( n(α 0 (F n, G n ) α) < σ α Φ 1 (β)) e 2n(α α0(f,g))2 If α 0 (F, G) < α and n(α α 0 (F, G)) 2 log 2 σ α Φ 1 (β), P F,G ( n(α 0 (F, G) α) σ α Φ 1 (β)) 2e ( σαφ 1 (β)+ n(α α 0(F,G))) 2 Eustasio del Barrio Testing approximate stochastic order 23 / 37

53 Inference for approximate stochastic order Testing for essential stochastic order Nonasymptotic bounds Theorem If α 0 (F, G) > α then P F,G ( n(α 0 (F n, G n ) α) < σ α Φ 1 (β)) e 2n(α α0(f,g))2 If α 0 (F, G) < α and n(α α 0 (F, G)) 2 log 2 σ α Φ 1 (β), P F,G ( n(α 0 (F, G) α) σ α Φ 1 (β)) 2e ( σαφ 1 (β)+ n(α α 0(F,G))) 2 Test is u.e.c. for H 0 : α 0 (F, G) α vs. H a : α 0 (F, G) < α if α < α < α Eustasio del Barrio Testing approximate stochastic order 23 / 37

54 Inference for approximate stochastic order Testing for essential stochastic order Nonasymptotic bounds Theorem If α 0 (F, G) > α then P F,G ( n(α 0 (F n, G n ) α) < σ α Φ 1 (β)) e 2n(α α0(f,g))2 If α 0 (F, G) < α and n(α α 0 (F, G)) 2 log 2 σ α Φ 1 (β), P F,G ( n(α 0 (F, G) α) σ α Φ 1 (β)) 2e ( σαφ 1 (β)+ n(α α 0(F,G))) 2 Test is u.e.c. for H 0 : α 0 (F, G) α vs. H a : α 0 (F, G) < α if α < α < α Compare to case H 0 : F st G vs. H a : F st G Eustasio del Barrio Testing approximate stochastic order 23 / 37

55 Inference for approximate stochastic order Testing for essential stochastic order Nonasymptotic bounds Theorem If α 0 (F, G) > α then P F,G ( n(α 0 (F n, G n ) α) < σ α Φ 1 (β)) e 2n(α α0(f,g))2 If α 0 (F, G) < α and n(α α 0 (F, G)) 2 log 2 σ α Φ 1 (β), P F,G ( n(α 0 (F, G) α) σ α Φ 1 (β)) 2e ( σαφ 1 (β)+ n(α α 0(F,G))) 2 Test is u.e.c. for H 0 : α 0 (F, G) α vs. H a : α 0 (F, G) < α if α < α < α Compare to case H 0 : F st G vs. H a : F st G Try to assess F st G up to α = 0.05 contamination; β = 0.05 Eustasio del Barrio Testing approximate stochastic order 23 / 37

56 Inference for approximate stochastic order Testing for essential stochastic order Nonasymptotic bounds Theorem If α 0 (F, G) > α then P F,G ( n(α 0 (F n, G n ) α) < σ α Φ 1 (β)) e 2n(α α0(f,g))2 If α 0 (F, G) < α and n(α α 0 (F, G)) 2 log 2 σ α Φ 1 (β), P F,G ( n(α 0 (F, G) α) σ α Φ 1 (β)) 2e ( σαφ 1 (β)+ n(α α 0(F,G))) 2 Test is u.e.c. for H 0 : α 0 (F, G) α vs. H a : α 0 (F, G) < α if α < α < α Compare to case H 0 : F st G vs. H a : F st G Try to assess F st G up to α = 0.05 contamination; β = 0.05 Want to detect alternatives with α 0 (F, G) 0.01 with power 0.9 Eustasio del Barrio Testing approximate stochastic order 23 / 37

57 Inference for approximate stochastic order Testing for essential stochastic order Nonasymptotic bounds Theorem If α 0 (F, G) > α then P F,G ( n(α 0 (F n, G n ) α) < σ α Φ 1 (β)) e 2n(α α0(f,g))2 If α 0 (F, G) < α and n(α α 0 (F, G)) 2 log 2 σ α Φ 1 (β), P F,G ( n(α 0 (F, G) α) σ α Φ 1 (β)) 2e ( σαφ 1 (β)+ n(α α 0(F,G))) 2 Test is u.e.c. for H 0 : α 0 (F, G) α vs. H a : α 0 (F, G) < α if α < α < α Compare to case H 0 : F st G vs. H a : F st G Try to assess F st G up to α = 0.05 contamination; β = 0.05 Want to detect alternatives with α 0 (F, G) 0.01 with power 0.9 Take n = 8143 Eustasio del Barrio Testing approximate stochastic order 23 / 37

58 Inference for approximate stochastic order Confidence bounds Confidence bounds Instead of testing for/against contaminated stochastic order, report upper/lower bounds for true contamination level, α 0 (F, G) For β < 1 2, α 0 (F n, G n ) nˆσ n Φ 1 (β) ˆσ 2 n = min t:fn(t) G n(t)=α 0(F n,g n)) σ 2 t, σ 2 t = t(1 t) + (t α 0 (F n, G n ))(1 t + α 0 (F n, G n )) is an upper bound with asymptotic confidence level at least 1 β Better use bias corrected α 0 (F n, G n ) BOOT α 0 (F n, G n ) nk 1 β (α 0 (F n, G n )) is a lower confidence bound for α 0 (F, G) with asymptotic confidence level 1 β Quantiles K 1 β (α 0 (F n, G n )) numerically approximated Eustasio del Barrio Testing approximate stochastic order 24 / 37

59 Inference for approximate stochastic order Paired sampling Dependent data Often X = pre-treatment, Y = post-treatment measurement (X, Y ) H with marginals F and G Has patient improved with treatment? F st G? As before, H 0 : F st G vs. H a : F st G not testable Consider instead H 0 : α 0 (F, G) α, vs. H a : α 0 (F, G) < α (X 1, Y 1 ),..., (X n, Y n ) i.i.d. random vectors with common joint d.f. H H(x, y) = C(F (x), G(y)), C copula α 0 (F, G) depends only on marginals; α 0 (F n, G n ) consistent estimator Distribution of α 0 (F n, G n ) does depend on C: n (α0 (F n, G n ) α 0 (F, G)) w sup B C (G(x), F (x)), x:g(x) F (x)=α 0(F,G) B C centered Gaussian on [0, 1] 2, covariance K C ((s, t), (s, t )) = s s + t t (s t)(s t ) C(t, s) C(t, s ) Eustasio del Barrio Testing approximate stochastic order 25 / 37

60 Inference for approximate stochastic order Paired sampling Testing for approximate stochastic order, dependent data Now α(1 α) Var(B C (t, t α)) 1 α 2 2 t 1+α 2, t [α, 1] equality for antimonotone coupling C(s, t) = (s + t 1) + K C,β (F, G) β-quantile of rhs in limit distribution; for β (0, 1 2 ) K C,β (F, G) (1 α 0 (F, G) 2 ) 1/2 Φ 1 (β) Similar to independent case, set σ 2 α = 1 α 2 ; reject α 0 (F, G) α if n(α0 (F n, G n ) α) < σ α Φ 1 (β) Eustasio del Barrio Testing approximate stochastic order 26 / 37

61 Inference for approximate stochastic order Paired sampling Testing for approximate stochastic order, dependent data As before, test uniformly asymptotically consistent: lim n sup [ P H n(α0 (F n, G n ) α) < σ α Φ 1 (β) ] H H 0 = lim P [ n H n(α0 (F n, G n ) α) < σ π0 Φ 1 (β) ] = β, H joint d.f. with marginals F 1 α 1+α 2 U(0, 2 ) + 1+α 1+α 2 U( 2, 1 + α(1 α) 2 ), G U(0, 1) and copula C(s, t) = (s + t 1) +. Eustasio del Barrio Testing approximate stochastic order 27 / 37

62 Inference for approximate stochastic order Paired sampling Nonasymptotic bounds: if α 0 (F, G) > α then P H [ n(α0 (F n, G n ) α < σ α Φ 1 (β) ] e n 2 (α α0(fn,gn)2, If α 0 (F, G) < α and n(α α 0 (F, G)) 2 2 log 2 σ α Φ 1 (β), P H [ n(α0 (F n, G n ) α) σ α Φ 1 (β) ] 2e n 2 [(α α0(f,g))+ 2 σα n Φ 1 (β)] 2. Eustasio del Barrio Testing approximate stochastic order 28 / 37

63 Inference for approximate stochastic order Paired sampling Nonasymptotic bounds: if α 0 (F, G) > α then P H [ n(α0 (F n, G n ) α < σ α Φ 1 (β) ] e n 2 (α α0(fn,gn)2, If α 0 (F, G) < α and n(α α 0 (F, G)) 2 2 log 2 σ α Φ 1 (β), P H [ n(α0 (F n, G n ) α) σ α Φ 1 (β) ] 2e n 2 [(α α0(f,g))+ 2 σα n Φ 1 (β)] 2. Independent vs. dependent setup In independent setup rejection of H 0 : α 0 (F, G) α when n(α0 (F n, G n ) α) < σα 2 Φ 1 (β). Under dependence, the extra 2 factor allows to control uniformly type I error probability Eustasio del Barrio Testing approximate stochastic order 28 / 37

64 Implementation, simulation & data example Implementation issues Testing for essential stochastic order: finite sample performance Consider again H 0 : α 0 (F, G) α, vs. H a : α 0 (F, G) < α Eustasio del Barrio Testing approximate stochastic order 29 / 37

65 Implementation, simulation & data example Implementation issues Testing for essential stochastic order: finite sample performance Consider again H 0 : α 0 (F, G) α, vs. H a : α 0 (F, G) < α Rejection of H 0 if n(α0 (F n, G n ) α) < σ α Φ 1 (β), σ 2 α = 1 α2 2, asympt. of level β; type I-type II error probs. exponentially 0 Eustasio del Barrio Testing approximate stochastic order 29 / 37

66 Implementation, simulation & data example Implementation issues Testing for essential stochastic order: finite sample performance Consider again H 0 : α 0 (F, G) α, vs. H a : α 0 (F, G) < α Rejection of H 0 if n(α0 (F n, G n ) α) < σ α Φ 1 (β), σ 2 α = 1 α2 2, asympt. of level β; type I-type II error probs. exponentially 0 σ α from worst case choice; possible improvement from estimated ˆσ n A more important improvement: E(α 0 (F n, G n )) α 0 (F, G); estimate bias by bias BOOT (α 0 (F n, G n )) := E (α 0 (F n, G n)) α 0 (F n, G n ) Eustasio del Barrio Testing approximate stochastic order 29 / 37

67 Implementation, simulation & data example Implementation issues Testing for essential stochastic order: finite sample performance Consider again H 0 : α 0 (F, G) α, vs. H a : α 0 (F, G) < α Rejection of H 0 if n(α0 (F n, G n ) α) < σ α Φ 1 (β), σ 2 α = 1 α2 2, asympt. of level β; type I-type II error probs. exponentially 0 σ α from worst case choice; possible improvement from estimated ˆσ n A more important improvement: E(α 0 (F n, G n )) α 0 (F, G); estimate bias by bias BOOT (α 0 (F n, G n )) := E (α 0 (F n, G n)) α 0 (F n, G n ) Define ˆα n,boot := α 0 (F n, G n ) bias BOOT (α 0 (F n, G n )). Eustasio del Barrio Testing approximate stochastic order 29 / 37

68 Implementation, simulation & data example Implementation issues Testing for essential stochastic order: finite sample performance Consider again H 0 : α 0 (F, G) α, vs. H a : α 0 (F, G) < α Rejection of H 0 if n(α0 (F n, G n ) α) < σ α Φ 1 (β), σ 2 α = 1 α2 2, asympt. of level β; type I-type II error probs. exponentially 0 σ α from worst case choice; possible improvement from estimated ˆσ n A more important improvement: E(α 0 (F n, G n )) α 0 (F, G); estimate bias by bias BOOT (α 0 (F n, G n )) := E (α 0 (F n, G n)) α 0 (F n, G n ) Define ˆα n,boot := α 0 (F n, G n ) bias BOOT (α 0 (F n, G n )). Reject H 0 if n(ˆαn,boot α) < ˆσ n Φ 1 (β), Eustasio del Barrio Testing approximate stochastic order 29 / 37

69 Implementation, simulation & data example Implementation issues Testing for essential stochastic order: finite sample performance Consider again H 0 : α 0 (F, G) α, vs. H a : α 0 (F, G) < α Rejection of H 0 if n(α0 (F n, G n ) α) < σ α Φ 1 (β), σ 2 α = 1 α2 2, asympt. of level β; type I-type II error probs. exponentially 0 σ α from worst case choice; possible improvement from estimated ˆσ n A more important improvement: E(α 0 (F n, G n )) α 0 (F, G); estimate bias by bias BOOT (α 0 (F n, G n )) := E (α 0 (F n, G n)) α 0 (F n, G n ) Define ˆα n,boot := α 0 (F n, G n ) bias BOOT (α 0 (F n, G n )). Reject H 0 if n(ˆαn,boot α) < ˆσ n Φ 1 (β), Test asympt. of level β Eustasio del Barrio Testing approximate stochastic order 29 / 37

70 Implementation, simulation & data example Simulation setup F α,a U(a, 1 + a); F α,b 1 α 1+α 2 U(0, G U(0, 1) 2 ) + 1+α 2 U( 1+α 2, 1 + α(1 α) 2 ) F α,a F α,b F α,a, G worst choice in test against essential s.o. F α,b, G worst choice in test for essential s.o. Eustasio del Barrio Testing approximate stochastic order 30 / 37

71 Implementation, simulation & data example Simulation results Testing for essential stochastic order Table : Observed rejection frequencies. H 0 : α 0 (F, G) α vs. H a : α 0 (F, G) < α G = U(0, 1), m = n; reject if n(α 0 (F n, G n ) α) < σ α Φ 1 (0.05) α n F 0.1,a F 0.1,b F 0.05,a F 0.05,b F 0.01,a F 0.01,b F Nonasymptotic estimate n = 8143 Eustasio del Barrio Testing approximate stochastic order 31 / 37

72 Implementation, simulation & data example Simulation results Testing for essential stochastic order Table : Observed rejection frequencies. H 0 : α 0 (F, G) α vs. H a : α 0 (F, G) < α G = U(0, 1), m = n; reject if n(ˆα n,boot α) < ˆσ n Φ 1 (0.05) α n F 0.1,a F 0.1,b F 0.05,a F 0.05,b F 0.01,a F 0.01,b F Eustasio del Barrio Testing approximate stochastic order 32 / 37

73 Implementation, simulation & data example Simulation results Testing against essential stochastic order Table : Observed rejection frequencies. H 0 : α 0 (F, G) α vs. H a : α 0 (F, G) > α G = U(0, 1), m = n; reject if n(α 0 (F n, G n ) α) > K 0.95 (α) α n F 0 F 0.01,b F 0.01,a F 0.05,b F 0.05,a F 0.1,b F 0.1,a Eustasio del Barrio Testing approximate stochastic order 33 / 37

74 Implementation, simulation & data example Simulation results Testing for essential stochastic order, dependent case Table : Observed rejection frequencies. H 0 : α 0 (F, G) α vs. H a : α 0 (F, G) < α G = U(0, 1), m = n; reject if n(ˆα 0 (F n, G n α) < ˆσ n Φ 1 (0.05) α n H 0.1,a H 0.1,b H 0.05,a H 0.05,b H 0.01,a H 0.01,b H H π,a independent marginals U(π, 1 + π), U(0, 1); H 0 = H 0,a H π,b marginals F 1 π 1+π 2 U(0, 2 ) + 1+π 1+π 2 U( 2, 1 + π(1 π) 2 ), G U(0, 1), copula C(s, t) = (s + t 1) + Eustasio del Barrio Testing approximate stochastic order 34 / 37

75 Implementation, simulation & data example Case study Data: National Health and Nutrition Examination Survey Evolution with age of the heights of boys and girls Sample sizes by age (boys, top) π(fn,gm) = π(gm,fn) = EDF's Gm Age 10 Boys Fn Age 10 Girls heights Eustasio del Barrio Testing approximate stochastic order 35 / 37

76 Implementation, simulation & data example Case study 95%-Upper bounds by age for α 0 (F a, G a ) (top row) and α 0 (G a, F a ) (bottom) Upper 95% confidence bounds for the stochastic dominance levels age Statistical evidence that girls are taller than boys at Eustasio del Barrio Testing approximate stochastic order 36 / 37

77 Conclusions Conclusions Trimmed stochastic order models capture adequately deviations from exact stochastic order Provided valid inference models/methods Valid testing procedures with controlled error probabilities Nonasymptotic bounds; uniformly exponentially consistent tests Good finite sample performance through bootstrap correction Eustasio del Barrio Testing approximate stochastic order 37 / 37

78 Conclusions Conclusions Trimmed stochastic order models capture adequately deviations from exact stochastic order Provided valid inference models/methods Valid testing procedures with controlled error probabilities Nonasymptotic bounds; uniformly exponentially consistent tests Good finite sample performance through bootstrap correction Thanks for your attention! Eustasio del Barrio Testing approximate stochastic order 37 / 37

Estimation of a Two-component Mixture Model

Estimation of a Two-component Mixture Model Estimation of a Two-component Mixture Model Bodhisattva Sen 1,2 University of Cambridge, Cambridge, UK Columbia University, New York, USA Indian Statistical Institute, Kolkata, India 6 August, 2012 1 Joint

More information

Asymptotic Statistics-VI. Changliang Zou

Asymptotic Statistics-VI. Changliang Zou Asymptotic Statistics-VI Changliang Zou Kolmogorov-Smirnov distance Example (Kolmogorov-Smirnov confidence intervals) We know given α (0, 1), there is a well-defined d = d α,n such that, for any continuous

More information

ORIGINS OF STOCHASTIC PROGRAMMING

ORIGINS OF STOCHASTIC PROGRAMMING ORIGINS OF STOCHASTIC PROGRAMMING Early 1950 s: in applications of Linear Programming unknown values of coefficients: demands, technological coefficients, yields, etc. QUOTATION Dantzig, Interfaces 20,1990

More information

Lecture 13: p-values and union intersection tests

Lecture 13: p-values and union intersection tests Lecture 13: p-values and union intersection tests p-values After a hypothesis test is done, one method of reporting the result is to report the size α of the test used to reject H 0 or accept H 0. If α

More information

Testing Downside-Risk Efficiency Under Distress

Testing Downside-Risk Efficiency Under Distress Testing Downside-Risk Efficiency Under Distress Jesus Gonzalo Universidad Carlos III de Madrid Jose Olmo City University of London XXXIII Simposio Analisis Economico 1 Some key lines Risk vs Uncertainty.

More information

Hypothesis Testing. 1 Definitions of test statistics. CB: chapter 8; section 10.3

Hypothesis Testing. 1 Definitions of test statistics. CB: chapter 8; section 10.3 Hypothesis Testing CB: chapter 8; section 0.3 Hypothesis: statement about an unknown population parameter Examples: The average age of males in Sweden is 7. (statement about population mean) The lowest

More information

Model Specification Testing in Nonparametric and Semiparametric Time Series Econometrics. Jiti Gao

Model Specification Testing in Nonparametric and Semiparametric Time Series Econometrics. Jiti Gao Model Specification Testing in Nonparametric and Semiparametric Time Series Econometrics Jiti Gao Department of Statistics School of Mathematics and Statistics The University of Western Australia Crawley

More information

Convergence of Multivariate Quantile Surfaces

Convergence of Multivariate Quantile Surfaces Convergence of Multivariate Quantile Surfaces Adil Ahidar Institut de Mathématiques de Toulouse - CERFACS August 30, 2013 Adil Ahidar (Institut de Mathématiques de Toulouse Convergence - CERFACS) of Multivariate

More information

OPTIMAL TRANSPORTATION PLANS AND CONVERGENCE IN DISTRIBUTION

OPTIMAL TRANSPORTATION PLANS AND CONVERGENCE IN DISTRIBUTION OPTIMAL TRANSPORTATION PLANS AND CONVERGENCE IN DISTRIBUTION J.A. Cuesta-Albertos 1, C. Matrán 2 and A. Tuero-Díaz 1 1 Departamento de Matemáticas, Estadística y Computación. Universidad de Cantabria.

More information

14.30 Introduction to Statistical Methods in Economics Spring 2009

14.30 Introduction to Statistical Methods in Economics Spring 2009 MIT OpenCourseWare http://ocw.mit.edu 4.0 Introduction to Statistical Methods in Economics Spring 009 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.

More information

Asymptotic Distributions for the Nelson-Aalen and Kaplan-Meier estimators and for test statistics.

Asymptotic Distributions for the Nelson-Aalen and Kaplan-Meier estimators and for test statistics. Asymptotic Distributions for the Nelson-Aalen and Kaplan-Meier estimators and for test statistics. Dragi Anevski Mathematical Sciences und University November 25, 21 1 Asymptotic distributions for statistical

More information

Random Variables and Their Distributions

Random Variables and Their Distributions Chapter 3 Random Variables and Their Distributions A random variable (r.v.) is a function that assigns one and only one numerical value to each simple event in an experiment. We will denote r.vs by capital

More information

STAT 135 Lab 5 Bootstrapping and Hypothesis Testing

STAT 135 Lab 5 Bootstrapping and Hypothesis Testing STAT 135 Lab 5 Bootstrapping and Hypothesis Testing Rebecca Barter March 2, 2015 The Bootstrap Bootstrap Suppose that we are interested in estimating a parameter θ from some population with members x 1,...,

More information

Stochastic Convergence, Delta Method & Moment Estimators

Stochastic Convergence, Delta Method & Moment Estimators Stochastic Convergence, Delta Method & Moment Estimators Seminar on Asymptotic Statistics Daniel Hoffmann University of Kaiserslautern Department of Mathematics February 13, 2015 Daniel Hoffmann (TU KL)

More information

Nonparametric Inference via Bootstrapping the Debiased Estimator

Nonparametric Inference via Bootstrapping the Debiased Estimator Nonparametric Inference via Bootstrapping the Debiased Estimator Yen-Chi Chen Department of Statistics, University of Washington ICSA-Canada Chapter Symposium 2017 1 / 21 Problem Setup Let X 1,, X n be

More information

Extreme Value Analysis and Spatial Extremes

Extreme Value Analysis and Spatial Extremes Extreme Value Analysis and Department of Statistics Purdue University 11/07/2013 Outline Motivation 1 Motivation 2 Extreme Value Theorem and 3 Bayesian Hierarchical Models Copula Models Max-stable Models

More information

Lecture 12 November 3

Lecture 12 November 3 STATS 300A: Theory of Statistics Fall 2015 Lecture 12 November 3 Lecturer: Lester Mackey Scribe: Jae Hyuck Park, Christian Fong Warning: These notes may contain factual and/or typographic errors. 12.1

More information

STAT 512 sp 2018 Summary Sheet

STAT 512 sp 2018 Summary Sheet STAT 5 sp 08 Summary Sheet Karl B. Gregory Spring 08. Transformations of a random variable Let X be a rv with support X and let g be a function mapping X to Y with inverse mapping g (A = {x X : g(x A}

More information

Math 181B Homework 1 Solution

Math 181B Homework 1 Solution Math 181B Homework 1 Solution 1. Write down the likelihood: L(λ = n λ X i e λ X i! (a One-sided test: H 0 : λ = 1 vs H 1 : λ = 0.1 The likelihood ratio: where LR = L(1 L(0.1 = 1 X i e n 1 = λ n X i e nλ

More information

The Bootstrap: Theory and Applications. Biing-Shen Kuo National Chengchi University

The Bootstrap: Theory and Applications. Biing-Shen Kuo National Chengchi University The Bootstrap: Theory and Applications Biing-Shen Kuo National Chengchi University Motivation: Poor Asymptotic Approximation Most of statistical inference relies on asymptotic theory. Motivation: Poor

More information

Lecture 9: October 25, Lower bounds for minimax rates via multiple hypotheses

Lecture 9: October 25, Lower bounds for minimax rates via multiple hypotheses Information and Coding Theory Autumn 07 Lecturer: Madhur Tulsiani Lecture 9: October 5, 07 Lower bounds for minimax rates via multiple hypotheses In this lecture, we extend the ideas from the previous

More information

LECTURE 3. Last time:

LECTURE 3. Last time: LECTURE 3 Last time: Mutual Information. Convexity and concavity Jensen s inequality Information Inequality Data processing theorem Fano s Inequality Lecture outline Stochastic processes, Entropy rate

More information

Lecture 8 Inequality Testing and Moment Inequality Models

Lecture 8 Inequality Testing and Moment Inequality Models Lecture 8 Inequality Testing and Moment Inequality Models Inequality Testing In the previous lecture, we discussed how to test the nonlinear hypothesis H 0 : h(θ 0 ) 0 when the sample information comes

More information

Lecture 3. Inference about multivariate normal distribution

Lecture 3. Inference about multivariate normal distribution Lecture 3. Inference about multivariate normal distribution 3.1 Point and Interval Estimation Let X 1,..., X n be i.i.d. N p (µ, Σ). We are interested in evaluation of the maximum likelihood estimates

More information

Stat 710: Mathematical Statistics Lecture 31

Stat 710: Mathematical Statistics Lecture 31 Stat 710: Mathematical Statistics Lecture 31 Jun Shao Department of Statistics University of Wisconsin Madison, WI 53706, USA Jun Shao (UW-Madison) Stat 710, Lecture 31 April 13, 2009 1 / 13 Lecture 31:

More information

ST5215: Advanced Statistical Theory

ST5215: Advanced Statistical Theory Department of Statistics & Applied Probability Wednesday, October 19, 2011 Lecture 17: UMVUE and the first method of derivation Estimable parameters Let ϑ be a parameter in the family P. If there exists

More information

Chapter 3: Unbiased Estimation Lecture 22: UMVUE and the method of using a sufficient and complete statistic

Chapter 3: Unbiased Estimation Lecture 22: UMVUE and the method of using a sufficient and complete statistic Chapter 3: Unbiased Estimation Lecture 22: UMVUE and the method of using a sufficient and complete statistic Unbiased estimation Unbiased or asymptotically unbiased estimation plays an important role in

More information

Inference For High Dimensional M-estimates. Fixed Design Results

Inference For High Dimensional M-estimates. Fixed Design Results : Fixed Design Results Lihua Lei Advisors: Peter J. Bickel, Michael I. Jordan joint work with Peter J. Bickel and Noureddine El Karoui Dec. 8, 2016 1/57 Table of Contents 1 Background 2 Main Results and

More information

Confidence Intervals for Low-dimensional Parameters with High-dimensional Data

Confidence Intervals for Low-dimensional Parameters with High-dimensional Data Confidence Intervals for Low-dimensional Parameters with High-dimensional Data Cun-Hui Zhang and Stephanie S. Zhang Rutgers University and Columbia University September 14, 2012 Outline Introduction Methodology

More information

Hypothesis Testing - Frequentist

Hypothesis Testing - Frequentist Frequentist Hypothesis Testing - Frequentist Compare two hypotheses to see which one better explains the data. Or, alternatively, what is the best way to separate events into two classes, those originating

More information

Recall the Basics of Hypothesis Testing

Recall the Basics of Hypothesis Testing Recall the Basics of Hypothesis Testing The level of significance α, (size of test) is defined as the probability of X falling in w (rejecting H 0 ) when H 0 is true: P(X w H 0 ) = α. H 0 TRUE H 1 TRUE

More information

Statistical inference on Lévy processes

Statistical inference on Lévy processes Alberto Coca Cabrero University of Cambridge - CCA Supervisors: Dr. Richard Nickl and Professor L.C.G.Rogers Funded by Fundación Mutua Madrileña and EPSRC MASDOC/CCA student workshop 2013 26th March Outline

More information

LARGE SAMPLE BEHAVIOR OF SOME WELL-KNOWN ROBUST ESTIMATORS UNDER LONG-RANGE DEPENDENCE

LARGE SAMPLE BEHAVIOR OF SOME WELL-KNOWN ROBUST ESTIMATORS UNDER LONG-RANGE DEPENDENCE LARGE SAMPLE BEHAVIOR OF SOME WELL-KNOWN ROBUST ESTIMATORS UNDER LONG-RANGE DEPENDENCE C. LÉVY-LEDUC, H. BOISTARD, E. MOULINES, M. S. TAQQU, AND V. A. REISEN Abstract. The paper concerns robust location

More information

Modelling Dependence with Copulas and Applications to Risk Management. Filip Lindskog, RiskLab, ETH Zürich

Modelling Dependence with Copulas and Applications to Risk Management. Filip Lindskog, RiskLab, ETH Zürich Modelling Dependence with Copulas and Applications to Risk Management Filip Lindskog, RiskLab, ETH Zürich 02-07-2000 Home page: http://www.math.ethz.ch/ lindskog E-mail: lindskog@math.ethz.ch RiskLab:

More information

Monte-Carlo MMD-MA, Université Paris-Dauphine. Xiaolu Tan

Monte-Carlo MMD-MA, Université Paris-Dauphine. Xiaolu Tan Monte-Carlo MMD-MA, Université Paris-Dauphine Xiaolu Tan tan@ceremade.dauphine.fr Septembre 2015 Contents 1 Introduction 1 1.1 The principle.................................. 1 1.2 The error analysis

More information

Advanced Statistics II: Non Parametric Tests

Advanced Statistics II: Non Parametric Tests Advanced Statistics II: Non Parametric Tests Aurélien Garivier ParisTech February 27, 2011 Outline Fitting a distribution Rank Tests for the comparison of two samples Two unrelated samples: Mann-Whitney

More information

General Linear Test of a General Linear Hypothesis. Copyright c 2012 Dan Nettleton (Iowa State University) Statistics / 35

General Linear Test of a General Linear Hypothesis. Copyright c 2012 Dan Nettleton (Iowa State University) Statistics / 35 General Linear Test of a General Linear Hypothesis Copyright c 2012 Dan Nettleton (Iowa State University) Statistics 611 1 / 35 Suppose the NTGMM holds so that y = Xβ + ε, where ε N(0, σ 2 I). opyright

More information

Stability of optimization problems with stochastic dominance constraints

Stability of optimization problems with stochastic dominance constraints Stability of optimization problems with stochastic dominance constraints D. Dentcheva and W. Römisch Stevens Institute of Technology, Hoboken Humboldt-University Berlin www.math.hu-berlin.de/~romisch SIAM

More information

Exercises in stochastic analysis

Exercises in stochastic analysis Exercises in stochastic analysis Franco Flandoli, Mario Maurelli, Dario Trevisan The exercises with a P are those which have been done totally or partially) in the previous lectures; the exercises with

More information

Optimal Transport and Wasserstein Distance

Optimal Transport and Wasserstein Distance Optimal Transport and Wasserstein Distance The Wasserstein distance which arises from the idea of optimal transport is being used more and more in Statistics and Machine Learning. In these notes we review

More information

Conditional Distributions

Conditional Distributions Conditional Distributions The goal is to provide a general definition of the conditional distribution of Y given X, when (X, Y ) are jointly distributed. Let F be a distribution function on R. Let G(,

More information

Mathematical Methods for Neurosciences. ENS - Master MVA Paris 6 - Master Maths-Bio ( )

Mathematical Methods for Neurosciences. ENS - Master MVA Paris 6 - Master Maths-Bio ( ) Mathematical Methods for Neurosciences. ENS - Master MVA Paris 6 - Master Maths-Bio (2014-2015) Etienne Tanré - Olivier Faugeras INRIA - Team Tosca October 22nd, 2014 E. Tanré (INRIA - Team Tosca) Mathematical

More information

Bayesian and frequentist inequality tests

Bayesian and frequentist inequality tests Bayesian and frequentist inequality tests David M. Kaplan Longhao Zhuo Department of Economics, University of Missouri July 1, 2016 Abstract Bayesian and frequentist criteria are fundamentally different,

More information

ABC methods for phase-type distributions with applications in insurance risk problems

ABC methods for phase-type distributions with applications in insurance risk problems ABC methods for phase-type with applications problems Concepcion Ausin, Department of Statistics, Universidad Carlos III de Madrid Joint work with: Pedro Galeano, Universidad Carlos III de Madrid Simon

More information

Lecture 32: Asymptotic confidence sets and likelihoods

Lecture 32: Asymptotic confidence sets and likelihoods Lecture 32: Asymptotic confidence sets and likelihoods Asymptotic criterion In some problems, especially in nonparametric problems, it is difficult to find a reasonable confidence set with a given confidence

More information

Inference for Identifiable Parameters in Partially Identified Econometric Models

Inference for Identifiable Parameters in Partially Identified Econometric Models Inference for Identifiable Parameters in Partially Identified Econometric Models Joseph P. Romano Department of Statistics Stanford University romano@stat.stanford.edu Azeem M. Shaikh Department of Economics

More information

Testing Hypothesis. Maura Mezzetti. Department of Economics and Finance Università Tor Vergata

Testing Hypothesis. Maura Mezzetti. Department of Economics and Finance Università Tor Vergata Maura Department of Economics and Finance Università Tor Vergata Hypothesis Testing Outline It is a mistake to confound strangeness with mystery Sherlock Holmes A Study in Scarlet Outline 1 The Power Function

More information

6 Markov Chain Monte Carlo (MCMC)

6 Markov Chain Monte Carlo (MCMC) 6 Markov Chain Monte Carlo (MCMC) The underlying idea in MCMC is to replace the iid samples of basic MC methods, with dependent samples from an ergodic Markov chain, whose limiting (stationary) distribution

More information

A Comparison of Robust Estimators Based on Two Types of Trimming

A Comparison of Robust Estimators Based on Two Types of Trimming Submitted to the Bernoulli A Comparison of Robust Estimators Based on Two Types of Trimming SUBHRA SANKAR DHAR 1, and PROBAL CHAUDHURI 1, 1 Theoretical Statistics and Mathematics Unit, Indian Statistical

More information

Practical conditions on Markov chains for weak convergence of tail empirical processes

Practical conditions on Markov chains for weak convergence of tail empirical processes Practical conditions on Markov chains for weak convergence of tail empirical processes Olivier Wintenberger University of Copenhagen and Paris VI Joint work with Rafa l Kulik and Philippe Soulier Toronto,

More information

High Breakdown Analogs of the Trimmed Mean

High Breakdown Analogs of the Trimmed Mean High Breakdown Analogs of the Trimmed Mean David J. Olive Southern Illinois University April 11, 2004 Abstract Two high breakdown estimators that are asymptotically equivalent to a sequence of trimmed

More information

STAT 200C: High-dimensional Statistics

STAT 200C: High-dimensional Statistics STAT 200C: High-dimensional Statistics Arash A. Amini May 30, 2018 1 / 57 Table of Contents 1 Sparse linear models Basis Pursuit and restricted null space property Sufficient conditions for RNS 2 / 57

More information

Bootstrapping high dimensional vector: interplay between dependence and dimensionality

Bootstrapping high dimensional vector: interplay between dependence and dimensionality Bootstrapping high dimensional vector: interplay between dependence and dimensionality Xianyang Zhang Joint work with Guang Cheng University of Missouri-Columbia LDHD: Transition Workshop, 2014 Xianyang

More information

6.1 Variational representation of f-divergences

6.1 Variational representation of f-divergences ECE598: Information-theoretic methods in high-dimensional statistics Spring 2016 Lecture 6: Variational representation, HCR and CR lower bounds Lecturer: Yihong Wu Scribe: Georgios Rovatsos, Feb 11, 2016

More information

Fall 2017 STAT 532 Homework Peter Hoff. 1. Let P be a probability measure on a collection of sets A.

Fall 2017 STAT 532 Homework Peter Hoff. 1. Let P be a probability measure on a collection of sets A. 1. Let P be a probability measure on a collection of sets A. (a) For each n N, let H n be a set in A such that H n H n+1. Show that P (H n ) monotonically converges to P ( k=1 H k) as n. (b) For each n

More information

Confidence regions for stochastic variational inequalities

Confidence regions for stochastic variational inequalities Confidence regions for stochastic variational inequalities Shu Lu Department of Statistics and Operations Research, University of North Carolina at Chapel Hill, 355 Hanes Building, CB#326, Chapel Hill,

More information

IMPROVING TWO RESULTS IN MULTIPLE TESTING

IMPROVING TWO RESULTS IN MULTIPLE TESTING IMPROVING TWO RESULTS IN MULTIPLE TESTING By Sanat K. Sarkar 1, Pranab K. Sen and Helmut Finner Temple University, University of North Carolina at Chapel Hill and University of Duesseldorf October 11,

More information

Spring 2012 Math 541B Exam 1

Spring 2012 Math 541B Exam 1 Spring 2012 Math 541B Exam 1 1. A sample of size n is drawn without replacement from an urn containing N balls, m of which are red and N m are black; the balls are otherwise indistinguishable. Let X denote

More information

Multiscale Adaptive Inference on Conditional Moment Inequalities

Multiscale Adaptive Inference on Conditional Moment Inequalities Multiscale Adaptive Inference on Conditional Moment Inequalities Timothy B. Armstrong 1 Hock Peng Chan 2 1 Yale University 2 National University of Singapore June 2013 Conditional moment inequality models

More information

Normal (Gaussian) distribution The normal distribution is often relevant because of the Central Limit Theorem (CLT):

Normal (Gaussian) distribution The normal distribution is often relevant because of the Central Limit Theorem (CLT): Lecture Three Normal theory null distributions Normal (Gaussian) distribution The normal distribution is often relevant because of the Central Limit Theorem (CLT): A random variable which is a sum of many

More information

Super-Gaussian directions of random vectors

Super-Gaussian directions of random vectors Weizmann Institute & Tel Aviv University IMU-INdAM Conference in Analysis, Tel Aviv, June 2017 Gaussian approximation Many distributions in R n, for large n, have approximately Gaussian marginals. Classical

More information

University of California San Diego and Stanford University and

University of California San Diego and Stanford University and First International Workshop on Functional and Operatorial Statistics. Toulouse, June 19-21, 2008 K-sample Subsampling Dimitris N. olitis andjoseph.romano University of California San Diego and Stanford

More information

Weak convergence. Amsterdam, 13 November Leiden University. Limit theorems. Shota Gugushvili. Generalities. Criteria

Weak convergence. Amsterdam, 13 November Leiden University. Limit theorems. Shota Gugushvili. Generalities. Criteria Weak Leiden University Amsterdam, 13 November 2013 Outline 1 2 3 4 5 6 7 Definition Definition Let µ, µ 1, µ 2,... be probability measures on (R, B). It is said that µ n converges weakly to µ, and we then

More information

Hypothesis Testing For Multilayer Network Data

Hypothesis Testing For Multilayer Network Data Hypothesis Testing For Multilayer Network Data Jun Li Dept of Mathematics and Statistics, Boston University Joint work with Eric Kolaczyk Outline Background and Motivation Geometric structure of multilayer

More information

40.530: Statistics. Professor Chen Zehua. Singapore University of Design and Technology

40.530: Statistics. Professor Chen Zehua. Singapore University of Design and Technology Singapore University of Design and Technology Lecture 9: Hypothesis testing, uniformly most powerful tests. The Neyman-Pearson framework Let P be the family of distributions of concern. The Neyman-Pearson

More information

Extremogram and Ex-Periodogram for heavy-tailed time series

Extremogram and Ex-Periodogram for heavy-tailed time series Extremogram and Ex-Periodogram for heavy-tailed time series 1 Thomas Mikosch University of Copenhagen Joint work with Richard A. Davis (Columbia) and Yuwei Zhao (Ulm) 1 Jussieu, April 9, 2014 1 2 Extremal

More information

Introduction to Empirical Processes and Semiparametric Inference Lecture 02: Overview Continued

Introduction to Empirical Processes and Semiparametric Inference Lecture 02: Overview Continued Introduction to Empirical Processes and Semiparametric Inference Lecture 02: Overview Continued Michael R. Kosorok, Ph.D. Professor and Chair of Biostatistics Professor of Statistics and Operations Research

More information

A Large-Sample Approach to Controlling the False Discovery Rate

A Large-Sample Approach to Controlling the False Discovery Rate A Large-Sample Approach to Controlling the False Discovery Rate Christopher R. Genovese Department of Statistics Carnegie Mellon University Larry Wasserman Department of Statistics Carnegie Mellon University

More information

TERENCE TAO S AN EPSILON OF ROOM CHAPTER 3 EXERCISES. 1. Exercise 1.3.1

TERENCE TAO S AN EPSILON OF ROOM CHAPTER 3 EXERCISES. 1. Exercise 1.3.1 TRNC TAO S AN PSILON OF ROOM CHAPTR 3 RCISS KLLR VANDBOGRT 1. xercise 1.3.1 We merely consider the inclusion f f, viewed as an element of L p (, χ, µ), where all nonmeasurable subnull sets are given measure

More information

Robust Backtesting Tests for Value-at-Risk Models

Robust Backtesting Tests for Value-at-Risk Models Robust Backtesting Tests for Value-at-Risk Models Jose Olmo City University London (joint work with Juan Carlos Escanciano, Indiana University) Far East and South Asia Meeting of the Econometric Society

More information

Non-parametric Inference and Resampling

Non-parametric Inference and Resampling Non-parametric Inference and Resampling Exercises by David Wozabal (Last update. Juni 010) 1 Basic Facts about Rank and Order Statistics 1.1 10 students were asked about the amount of time they spend surfing

More information

Approximating the gains from trade in large markets

Approximating the gains from trade in large markets Approximating the gains from trade in large markets Ellen Muir Supervisors: Peter Taylor & Simon Loertscher Joint work with Kostya Borovkov The University of Melbourne 7th April 2015 Outline Setup market

More information

The Contraction Method on C([0, 1]) and Donsker s Theorem

The Contraction Method on C([0, 1]) and Donsker s Theorem The Contraction Method on C([0, 1]) and Donsker s Theorem Henning Sulzbach J. W. Goethe-Universität Frankfurt a. M. YEP VII Probability, random trees and algorithms Eindhoven, March 12, 2010 joint work

More information

STAT 830 Non-parametric Inference Basics

STAT 830 Non-parametric Inference Basics STAT 830 Non-parametric Inference Basics Richard Lockhart Simon Fraser University STAT 801=830 Fall 2012 Richard Lockhart (Simon Fraser University)STAT 830 Non-parametric Inference Basics STAT 801=830

More information

Optimality conditions for unconstrained optimization. Outline

Optimality conditions for unconstrained optimization. Outline Optimality conditions for unconstrained optimization Daniel P. Robinson Department of Applied Mathematics and Statistics Johns Hopkins University September 13, 2018 Outline 1 The problem and definitions

More information

2WB05 Simulation Lecture 7: Output analysis

2WB05 Simulation Lecture 7: Output analysis 2WB05 Simulation Lecture 7: Output analysis Marko Boon http://www.win.tue.nl/courses/2wb05 December 17, 2012 Outline 2/33 Output analysis of a simulation Confidence intervals Warm-up interval Common random

More information

Extremogram and ex-periodogram for heavy-tailed time series

Extremogram and ex-periodogram for heavy-tailed time series Extremogram and ex-periodogram for heavy-tailed time series 1 Thomas Mikosch University of Copenhagen Joint work with Richard A. Davis (Columbia) and Yuwei Zhao (Ulm) 1 Zagreb, June 6, 2014 1 2 Extremal

More information

ON TWO RESULTS IN MULTIPLE TESTING

ON TWO RESULTS IN MULTIPLE TESTING ON TWO RESULTS IN MULTIPLE TESTING By Sanat K. Sarkar 1, Pranab K. Sen and Helmut Finner Temple University, University of North Carolina at Chapel Hill and University of Duesseldorf Two known results in

More information

Guarding against Spurious Discoveries in High Dimension. Jianqing Fan

Guarding against Spurious Discoveries in High Dimension. Jianqing Fan in High Dimension Jianqing Fan Princeton University with Wen-Xin Zhou September 30, 2016 Outline 1 Introduction 2 Spurious correlation and random geometry 3 Goodness Of Spurious Fit (GOSF) 4 Asymptotic

More information

arxiv: v1 [math.pr] 24 Aug 2018

arxiv: v1 [math.pr] 24 Aug 2018 On a class of norms generated by nonnegative integrable distributions arxiv:180808200v1 [mathpr] 24 Aug 2018 Michael Falk a and Gilles Stupfler b a Institute of Mathematics, University of Würzburg, Würzburg,

More information

Statistical Applications of Over-fitting due to Trimmings

Statistical Applications of Over-fitting due to Trimmings Statistical Applications of Over-fitting due to Trimmings Pedro C. Alvarez (joint work with E. del Barrio, J.A. Cuesta-Albertos and C. Matrán) 5èmes Rencontre de Statistiques Mathématiques BoSanTouVal09,

More information

THEOREMS, ETC., FOR MATH 515

THEOREMS, ETC., FOR MATH 515 THEOREMS, ETC., FOR MATH 515 Proposition 1 (=comment on page 17). If A is an algebra, then any finite union or finite intersection of sets in A is also in A. Proposition 2 (=Proposition 1.1). For every

More information

STAT 135 Lab 6 Duality of Hypothesis Testing and Confidence Intervals, GLRT, Pearson χ 2 Tests and Q-Q plots. March 8, 2015

STAT 135 Lab 6 Duality of Hypothesis Testing and Confidence Intervals, GLRT, Pearson χ 2 Tests and Q-Q plots. March 8, 2015 STAT 135 Lab 6 Duality of Hypothesis Testing and Confidence Intervals, GLRT, Pearson χ 2 Tests and Q-Q plots March 8, 2015 The duality between CI and hypothesis testing The duality between CI and hypothesis

More information

Probability and Measure

Probability and Measure Probability and Measure Robert L. Wolpert Institute of Statistics and Decision Sciences Duke University, Durham, NC, USA Convergence of Random Variables 1. Convergence Concepts 1.1. Convergence of Real

More information

Econometrica, Vol. 71, No. 1 (January, 2003), CONSISTENT TESTS FOR STOCHASTIC DOMINANCE. By Garry F. Barrett and Stephen G.

Econometrica, Vol. 71, No. 1 (January, 2003), CONSISTENT TESTS FOR STOCHASTIC DOMINANCE. By Garry F. Barrett and Stephen G. Econometrica, Vol. 71, No. 1 January, 2003), 71 104 CONSISTENT TESTS FOR STOCHASTIC DOMINANCE By Garry F. Barrett and Stephen G. Donald 1 Methods are proposed for testing stochastic dominance of any pre-specified

More information

Asymmetric least squares estimation and testing

Asymmetric least squares estimation and testing Asymmetric least squares estimation and testing Whitney Newey and James Powell Princeton University and University of Wisconsin-Madison January 27, 2012 Outline ALS estimators Large sample properties Asymptotic

More information

BIO5312 Biostatistics Lecture 13: Maximum Likelihood Estimation

BIO5312 Biostatistics Lecture 13: Maximum Likelihood Estimation BIO5312 Biostatistics Lecture 13: Maximum Likelihood Estimation Yujin Chung November 29th, 2016 Fall 2016 Yujin Chung Lec13: MLE Fall 2016 1/24 Previous Parametric tests Mean comparisons (normality assumption)

More information

The Adequate Bootstrap

The Adequate Bootstrap The Adequate Bootstrap arxiv:1608.05913v1 [stat.me] 21 Aug 2016 Toby Kenney Department of Mathematics and Statistics, Dalhousie University and Hong Gu Department of Mathematics and Statistics, Dalhousie

More information

2.6.3 Generalized likelihood ratio tests

2.6.3 Generalized likelihood ratio tests 26 HYPOTHESIS TESTING 113 263 Generalized likelihood ratio tests When a UMP test does not exist, we usually use a generalized likelihood ratio test to verify H 0 : θ Θ against H 1 : θ Θ\Θ It can be used

More information

High Dimensional Empirical Likelihood for Generalized Estimating Equations with Dependent Data

High Dimensional Empirical Likelihood for Generalized Estimating Equations with Dependent Data High Dimensional Empirical Likelihood for Generalized Estimating Equations with Dependent Data Song Xi CHEN Guanghua School of Management and Center for Statistical Science, Peking University Department

More information

Modelling Dependent Credit Risks

Modelling Dependent Credit Risks Modelling Dependent Credit Risks Filip Lindskog, RiskLab, ETH Zürich 30 November 2000 Home page:http://www.math.ethz.ch/ lindskog E-mail:lindskog@math.ethz.ch RiskLab:http://www.risklab.ch Modelling Dependent

More information

Mathematics Qualifying Examination January 2015 STAT Mathematical Statistics

Mathematics Qualifying Examination January 2015 STAT Mathematical Statistics Mathematics Qualifying Examination January 2015 STAT 52800 - Mathematical Statistics NOTE: Answer all questions completely and justify your derivations and steps. A calculator and statistical tables (normal,

More information

ECE 4400:693 - Information Theory

ECE 4400:693 - Information Theory ECE 4400:693 - Information Theory Dr. Nghi Tran Lecture 8: Differential Entropy Dr. Nghi Tran (ECE-University of Akron) ECE 4400:693 Lecture 1 / 43 Outline 1 Review: Entropy of discrete RVs 2 Differential

More information

Paul Schrimpf. January 23, UBC Economics 326. Statistics and Inference. Paul Schrimpf. Properties of estimators. Finite sample inference

Paul Schrimpf. January 23, UBC Economics 326. Statistics and Inference. Paul Schrimpf. Properties of estimators. Finite sample inference UBC Economics 326 January 23, 2018 1 2 3 4 Wooldridge (2013) appendix C Stock and Watson (2009) chapter 3 Angrist and Pischke (2014) chapter 1 appendix Diez, Barr, and Cetinkaya-Rundel (2012) chapters

More information

Final Exam. 1. (6 points) True/False. Please read the statements carefully, as no partial credit will be given.

Final Exam. 1. (6 points) True/False. Please read the statements carefully, as no partial credit will be given. 1. (6 points) True/False. Please read the statements carefully, as no partial credit will be given. (a) If X and Y are independent, Corr(X, Y ) = 0. (b) (c) (d) (e) A consistent estimator must be asymptotically

More information

ISyE 6644 Fall 2014 Test 3 Solutions

ISyE 6644 Fall 2014 Test 3 Solutions 1 NAME ISyE 6644 Fall 14 Test 3 Solutions revised 8/4/18 You have 1 minutes for this test. You are allowed three cheat sheets. Circle all final answers. Good luck! 1. [4 points] Suppose that the joint

More information

Chapter 5. Chapter 5 sections

Chapter 5. Chapter 5 sections 1 / 43 sections Discrete univariate distributions: 5.2 Bernoulli and Binomial distributions Just skim 5.3 Hypergeometric distributions 5.4 Poisson distributions Just skim 5.5 Negative Binomial distributions

More information

Vector Autoregressive Model. Vector Autoregressions II. Estimation of Vector Autoregressions II. Estimation of Vector Autoregressions I.

Vector Autoregressive Model. Vector Autoregressions II. Estimation of Vector Autoregressions II. Estimation of Vector Autoregressions I. Vector Autoregressive Model Vector Autoregressions II Empirical Macroeconomics - Lect 2 Dr. Ana Beatriz Galvao Queen Mary University of London January 2012 A VAR(p) model of the m 1 vector of time series

More information

2. Variance and Covariance: We will now derive some classic properties of variance and covariance. Assume real-valued random variables X and Y.

2. Variance and Covariance: We will now derive some classic properties of variance and covariance. Assume real-valued random variables X and Y. CS450 Final Review Problems Fall 08 Solutions or worked answers provided Problems -6 are based on the midterm review Identical problems are marked recap] Please consult previous recitations and textbook

More information

Chapter 4. Theory of Tests. 4.1 Introduction

Chapter 4. Theory of Tests. 4.1 Introduction Chapter 4 Theory of Tests 4.1 Introduction Parametric model: (X, B X, P θ ), P θ P = {P θ θ Θ} where Θ = H 0 +H 1 X = K +A : K: critical region = rejection region / A: acceptance region A decision rule

More information