Bayesian Test for Lifetime Performance Index of Exponential Distribution under Symmetric Entropy Loss Function

Similar documents
Chapter (8) Estimation and Confedence Intervals Examples

Statistics 3858 : Likelihood Ratio for Exponential Distribution

APPENDIX: STATISTICAL TOOLS

Journal of Modern Applied Statistical Methods

FUZZY ACCEPTANCE SAMPLING AND CHARACTERISTIC CURVES

Discrete Fourier Transform (DFT)

Technical Support Document Bias of the Minimum Statistic

NEW VERSION OF SZEGED INDEX AND ITS COMPUTATION FOR SOME NANOTUBES

The Interplay between l-max, l-min, p-max and p-min Stable Distributions

A Novel Approach to Recovering Depth from Defocus

Performance Rating of the Type 1 Half Logistic Gompertz Distribution: An Analytical Approach

Page 1. Before-After Control-Impact (BACI) Power Analysis For Several Related Populations (Variance Known) Richard A. Hinrichsen. September 24, 2010

DETECTION OF RELIABLE SOFTWARE USING SPRT ON TIME DOMAIN DATA

1985 AP Calculus BC: Section I

On the approximation of the constant of Napier

International Journal of Advanced and Applied Sciences

Probability & Statistics,

INTRODUCTION TO SAMPLING DISTRIBUTIONS

Session : Plasmas in Equilibrium

Chapter 2 Quality-Yield Measure for Very Low Fraction Defective

z 1+ 3 z = Π n =1 z f() z = n e - z = ( 1-z) e z e n z z 1- n = ( 1-z/2) 1+ 2n z e 2n e n -1 ( 1-z )/2 e 2n-1 1-2n -1 1 () z

Solution of Assignment #2

Hadamard Exponential Hankel Matrix, Its Eigenvalues and Some Norms

Bayesian Estimations in Insurance Theory and Practice

Chapter 2 Infinite Series Page 1 of 11. Chapter 2 : Infinite Series

PURE MATHEMATICS A-LEVEL PAPER 1

Restricted Factorial And A Remark On The Reduced Residue Classes

ln x = n e = 20 (nearest integer)

Comparison of Simple Indicator Kriging, DMPE, Full MV Approach for Categorical Random Variable Simulation

DTFT Properties. Example - Determine the DTFT Y ( e ) of n. Let. We can therefore write. From Table 3.1, the DTFT of x[n] is given by 1

Bayesian Test for Lifetime Performance Index of Ailamujia Distribution Under Squared Error Loss Function

Chapter Taylor Theorem Revisited

Character sums over generalized Lehmer numbers

Page 1 BACI. Before-After-Control-Impact Power Analysis For Several Related Populations (Variance Known) October 10, Richard A.

2617 Mark Scheme June 2005 Mark Scheme 2617 June 2005

THREE-WAY ROC ANALYSIS USING SAS SOFTWARE

On a problem of J. de Graaf connected with algebras of unbounded operators de Bruijn, N.G.

Empirical Study in Finite Correlation Coefficient in Two Phase Estimation

Markov s s & Chebyshev s Inequalities. Chebyshev s Theorem. Coefficient of Variation an example. Coefficient of Variation

A Note on Quantile Coupling Inequalities and Their Applications

The calculation method for SNE

Bipolar Junction Transistors

Folding of Hyperbolic Manifolds

Triple Play: From De Morgan to Stirling To Euler to Maclaurin to Stirling

The Exponential-Generalized Truncated Geometric (EGTG) Distribution: A New Lifetime Distribution

Worksheet: Taylor Series, Lagrange Error Bound ilearnmath.net

SOME IDENTITIES FOR THE GENERALIZED POLY-GENOCCHI POLYNOMIALS WITH THE PARAMETERS A, B AND C

POSTERIOR ESTIMATES OF TWO PARAMETER EXPONENTIAL DISTRIBUTION USING S-PLUS SOFTWARE

Review Exercises. 1. Evaluate using the definition of the definite integral as a Riemann Sum. Does the answer represent an area? 2

Option 3. b) xe dx = and therefore the series is convergent. 12 a) Divergent b) Convergent Proof 15 For. p = 1 1so the series diverges.

Digital Signal Processing, Fall 2006

New Sixteenth-Order Derivative-Free Methods for Solving Nonlinear Equations

2.29 Numerical Fluid Mechanics Spring 2015 Lecture 12

Derivation of a Predictor of Combination #1 and the MSE for a Predictor of a Position in Two Stage Sampling with Response Error.

They must have different numbers of electrons orbiting their nuclei. They must have the same number of neutrons in their nuclei.

Law of large numbers

COMPUTING FOLRIER AND LAPLACE TRANSFORMS. Sven-Ake Gustafson. be a real-valued func'cion, defined for nonnegative arguments.

Homotopy perturbation technique

Chapter 11.00C Physical Problem for Fast Fourier Transform Civil Engineering

A Strain-based Non-linear Elastic Model for Geomaterials

Evaluating Reliability Systems by Using Weibull & New Weibull Extension Distributions Mushtak A.K. Shiker

5.1 The Nuclear Atom

LECTURE 13 Filling the bands. Occupancy of Available Energy Levels

4.2 Design of Sections for Flexure

A NEW FAMILY OF GENERALIZED GAMMA DISTRIBUTION AND ITS APPLICATION

Direction: This test is worth 150 points. You are required to complete this test within 55 minutes.

Global Chaos Synchronization of the Hyperchaotic Qi Systems by Sliding Mode Control

Discrete Fourier Transform. Discrete Fourier Transform. Discrete Fourier Transform. Discrete Fourier Transform. Discrete Fourier Transform

A Prey-Predator Model with an Alternative Food for the Predator, Harvesting of Both the Species and with A Gestation Period for Interaction

Two Products Manufacturer s Production Decisions with Carbon Constraint

SECTION where P (cos θ, sin θ) and Q(cos θ, sin θ) are polynomials in cos θ and sin θ, provided Q is never equal to zero.

NET/JRF, GATE, IIT JAM, JEST, TIFR

UNIT 2: MATHEMATICAL ENVIRONMENT

10. Joint Moments and Joint Characteristic Functions

Application of Vague Soft Sets in students evaluation

CIVE322 BASIC HYDROLOGY Hydrologic Science and Engineering Civil and Environmental Engineering Department Fort Collins, CO (970)

H2 Mathematics Arithmetic & Geometric Series ( )

A Simple Proof that e is Irrational

MILLIKAN OIL DROP EXPERIMENT

STIRLING'S 1 FORMULA AND ITS APPLICATION

Solution to 1223 The Evil Warden.

ph People Grade Level: basic Duration: minutes Setting: classroom or field site

On the Reformulated Zagreb Indices of Certain Nanostructures

Electronic Supplementary Information

Search sequence databases 3 10/25/2016

Cramér-Rao Inequality: Let f(x; θ) be a probability density function with continuous parameter

6. Comparison of NLMS-OCF with Existing Algorithms

Estimation of Consumer Demand Functions When the Observed Prices Are the Same for All Sample Units

A NEW CURRENT TRANSFORMER SATURATION DETECTION ALGORITHM

MONTGOMERY COLLEGE Department of Mathematics Rockville Campus. 6x dx a. b. cos 2x dx ( ) 7. arctan x dx e. cos 2x dx. 2 cos3x dx

Iterative Methods of Order Four for Solving Nonlinear Equations

Direction: This test is worth 250 points. You are required to complete this test within 50 minutes.

Available online at Energy Procedia 4 (2011) Energy Procedia 00 (2010) GHGT-10

Data Assimilation 1. Alan O Neill National Centre for Earth Observation UK

How many neutrino species?

Part B: Transform Methods. Professor E. Ambikairajah UNSW, Australia

ARIMA Methods of Detecting Outliers in Time Series Periodic Processes

Figure 2-18 Thevenin Equivalent Circuit of a Noisy Resistor

Theoretical Analysis of Cross-Validation for Estimating the Risk of the k-nearest Neighbor Classifier

Chapter 13 GMM for Linear Factor Models in Discount Factor form. GMM on the pricing errors gives a crosssectional

Transcription:

Mathmatics ttrs 08; 4(): 0-4 http://www.scicpublishiggroup.com/j/ml doi: 0.648/j.ml.08040.5 ISSN: 575-503X (Prit); ISSN: 575-5056 (Oli) aysia Tst for iftim Prformac Idx of Expotial Distributio udr Symmtric Etropy oss Fuctio Guobig Fa School of Mathmatics ad Statistics, Hua Uivrsity of Fiac ad Ecoomics, Chagsha, Chia Email addrss: To cit this articl: Guobig Fa. aysia Tst for iftim Prformac Idx of Expotial Distributio udr Symmtric Etropy oss Fuctio. Mathmatics ttrs. Vol. 4, No., 08, pp. 0-4. doi: 0.648/j.ml.08040.5 Rcivd: March, 08; Accptd: March 7, 08; Publishd: May 4, 08 Abstract: Th task of this papr is to stimat th liftim prformac idx of Expotial distributio. A aysia tst procdur is stablishd udr symmtric tropy loss fuctio. Firstly, aysia stimatio of lif prformac idx is obtaid, th a aysia tst procdur for liftim prformac idx is proposd. Fially, a applid xampl is usd to illustrat th ffctivss of th proposd tst mthod. Kywords: iftim Prformac Idx, aysia Tst, Expotial Distributio, Symmtric Etropy oss Fuctio. Itroductio Procss capability idics (PCIs) hav b itroducd for masurig procss rproductio capability of a maufacturig idustry []. Thy ar importat i th maufacturig idustry to masur procss pottial ad prformac []. Thr ar may wll-kow PCIs hav b put forward. For xampl, th first procss capability idx CP is first itroducd by Jura i rfrc [3]. Th othr PCIs: C pk, C pm, C pmk ar proposd rspctivly [4-6]. Th study ad applicatios of PCIs hav draw grat atttio. For xampl, Abdolshah t al. [7] studid th applicatio of Cpmk i th fuzzy cotxt ad compar it with othr fuzzy PCIs. Kuriati [8] proposd a variabls rsubmittd samplig schm for o-sidd spcificatio limit basd o o-sidd PCIs. Sifi ad Nzhad [9] dvlopd a variabl samplig pla for rsubmittd lots basd o PCI ad aysia approach. To assss th products with th largr-th-bttr typ charactristics, Motgomry [0] proposd a spcial uilatral spcificatio PCI, amd as liftim prformac idx C, which is dfid as follows C µ = () σ whr is th lowr boud of th spcificatios. Th statistical ifrc of liftim prformac idx for products whos liftim distributd diffrt distributios hav b widly studid i rct yars. For xampl, t al. [] basd o typ II right csord sampl discussd th computatioal procdur of prformac assssmt of liftim idx of ormal products i th fuzzy cotxt. iu ad R [] discussd th aysia poit stimatio ad tst of liftim prformac idx wh th products liftim is distributd with xpotial distributio udr progrssivly typ II csord sampls. Wu t al. [3] dsigd accptac-samplig plas for a xpotial populatio basd o a liftim-prformac idx by miimiz th umbr of failurs rquird durig ispctio. Solima [4] discussd th aysia ad o-aysia stimators for liftim prformac idx basd o progrssiv first-failur csorig sampls arisig from th xpotiatd Frcht distributio. i [5] discussd th aysia stimatio ad tst procdur of liftim prformac idx for Ailamujia distributio udr squard rror loss fuctio. Suppos that X is th products liftim, ad it distributd with xpotial distributio whos probability dsity fuctio (pdf) ad cumulativ distributio fuctio (cdf) rspctivly: x f( x θ) = θ θ, x > 0 ()

Mathmatics ttrs 08; 4(): 0-4 x F( x; θ) = θ θ, x > 0 (3) Hr, θ > 0 is th ukow paramtr. Th aim of this papr is to study th aysia statistical ifrc of liftim prformac idx C. Th aysia stimatio ad th aysia tst procdur of C will b drivd. Th orgaizatio of this articl is as follows. Sctio will rcall som proprtis of th liftim prformac idx of product with th xpotial distributio. Th rlatioship btw C ad coformig rat will b also discussd. Furthrmor, th aysia stimator of C basd o th cojugat Gamma prior distributio is also obtaid udr symmtric tropy loss fuctio i Sctio 3. A w aysia hypothsis tstig procdur of liftim prformac idx is dvlopd i Sctio 4, ad a applicatio xampl is giv i Sctio 5. Fially, coclusios ar giv i Sctio 6.. iftim Prformac Idx Suppos that X is th liftim of such a product whos liftim distributio is xpotial distributio with pdf (). Th th procss ma µ = EX = / θ ad th procss stadard dviatio σ = Var (X) = / θ. Th th paramtr θ is oft calld th ma tim. Th th liftim prformac idx C of xpotial product ca b rwritt as follows C µ / θ = = = θ (4) σ / θ Th failur rat fuctio r( x) ca b drivd as f(x θ) θxp( θx) r (x) = = = θ (5) F(x θ) xp( θx) From (4) ad (5), w ca s that th largr th ma / θ is, th smallr th failur rat ad largr th liftim prformac idx C ar. Thrfor, th liftim prformac idx C ca rasoably ad accuratly rprst th product prformac of w products. Morovr, if th liftim X of a product xcds th lowr spcificatio limit, th th product is dfid as a coformig rat, ad ca b dfid as Pr P(X ) xp( x)dx = = θ θ θ C = =, < C < (6) Accordig to th quatio (6), w ca s thr xists a strictly icrasig rlatioship btw coformig rat P r ad th liftim prformac idx C. Tabl lists valus of C ad th corrspodig coformig rat P r. Wh th valus of C which ar ot listd i Tabl, th coformig rat P r ca b drivd through itrpolatio. Tabl. Th liftim prformac idx C vrsus th coformig rat P r. 0.00000 0.5 0.4686 0.575 0.65377 -.75 0.035 0.75 0.4383 0.65 0.6879 -.50 0.0300 0.00 0.44933 0.650 0.70469 -.00 0.04979 0.50 0.4737 0.700 0.7408 -.50 0.0809 0.300 0.49659 0.750 0.77880 -.5 0.0540 0.35 0.5096 0.775 0.7985 -.00 0.3534 0.350 0.505 0.800 0.8873-0.75 0.7377 0.375 0.5356 0.85 0.83946-0.50 0.33 0.400 0.5488 0.850 0.8607-0.5 0.8650 0.45 0.5670 0.875 0.8850 0.000 0.36788 0.450 0.57695 0.900 0.90484 0.05 0.3779 0.475 0.5956 0.95 0.9774 0.050 0.38674 0.500 0.60653 0.950 0.953 0.075 0.39657 0.55 0.689 0.975 0.9753 Th coformig rat ca b calculatd by dividig th umbr of coformig products by total umbr of products sampld. To accuratly stimatig rat, Motgomry [0] suggstd that th sampl siz must b ough larg. Howvr, a larg umbr of sampls ar usually ot practical from th poit of cost. I additio, a larg sampl is also ot practical du to tim limitatio or othr rstrictios such as matrial rsourcs, mchaical or xprimtal difficultis ad so o. Sic a o-to-o mappig xists btw th coformig rat P r ad th liftim prformac idx C. Thrfor, liftim prformac idx ca b ot oly a flxibl ad ffctiv tool to valuat th products prformac, but also b a ffctiv tool to stimat th coformig rat P r. 3. Estimatio 3.. Maximum iklihood Estimatio t X, X,, X b a liftim of sampl from th xpotial distributio with pdf (), x = ( x, x,, x ) is th obsrvatio of X = ( X, X,, X ) ad t = xi is th obsrvatio of statistic T = Xi. Th th liklihood fuctio corrspodig to pdf () is giv by (x; θ) = θt f(x θ) = θxp( θxi) = θ (7) Th maximum liklihood stimator of θ ca b asily dl [ ( θ; x)] drivd from th log-liklihood quatio = 0 as dθ follows: ˆM θ = (8) T W ca also asily prov that T is a radom variabl distributd with Gamma distributio Γ(, θ) with th

Guobig Fa: aysia Tst for iftim Prformac Idx of Expotial Distributio udr Symmtric Etropy oss Fuctio followig probability dsity fuctio: 3.. ays Estimatio θ θt ft( t; θ) = t, t > 0, θ > 0 Γ( ) This sctio will discuss th aysiaia stimatio of th liftim prformac idx of xpotial distributio with pdf () udr symmtric tropy loss fuctio (Zhao [6] ad i [7]) (9) ˆ θ θ ( ˆ, θ θ) = + (0) θ ˆ θ W ar itrstd i stimatig θ udr th symmtric tropy loss fuctio (0). mma t X, X,, X b a liftim of sampl from th xpotial distributio with pdf (), x = ( x, x,, x ) is th obsrvatio of X = ( X, X,, X ), Th udr th symmtric tropy loss fuctio (0), th uiqu aysia stimator of θ, say θˆ, is giv by ˆ E( θ X) θ = E( θ X) / () Proof. Obviously, th symmtric loss i () is strictly covx, th th uiqu aysia stimator is obtaid from th rlatio de[ ( ˆ θ, θ) X] = 0 dθ which rducs to (). Assum that th prior distributio of th paramtr θ is th Gamma prior distributio Γ (, β), with pdf β π θ β θ β θ Γ( ) βθ ( ;, ) =,, > 0, > 0 () Th combiig th liklihood fuctio (7) with th prior pdf (), th postrior pdf of θ ca b drivd usig ays Thorm as follows That is h( θ x) l( θ; x) π( θ) θ θ β θ Γ ( ) θt βθ + ( β + t) θ (3) θ X ~ Γ ( +, β + T). (4) Th udr th symmtric tropy loss fuctio (0), th aysia stimator of θ ca b drivd as follows + / ˆ E( θ X) β + T θ = = E( θ X) β + T + = ( β + T) ( + )( + ) Furthr th aysia stimator of C is Cˆ = ˆ θ β + T = ( + )( + ) / (5) (6) 4. aysia Tst of iftim Prformac Idx This sctio will propos a aysia tstig procdur to assss whthr th liftim prformac idx adhrs to th rquird lvl. Assum that th rquird idx valu of liftim prformac is largr tha c, whr c is th targt valu. First, w costruct th followig hypothsis: H : 0 C c H : C > c. (7) t Y= θβ ( + T) X, th for giv Sigificac lvl γ, w ca asily show that θ X ~ Γ ( +, β + T). t b th γ χ γ(( + )) prctil of χ (( + )). Th That is P( θβ ( + T) χ (( + )) X) = γ (8) γ χ γ(( + )) P( θ X) = γ ( β+ T) χ γ(( + )) P( θ X) = γ ( β+ T) (9) (0) Th th lowr cofidc limit of liftim prformac idx C with lvl γ ca b obtaid, as follows: χ (( )) ( ˆ γ + = C) ( + ) () Th th w proposd aysia tstig procdur of C is as follows: (i) For giv sampl siz, dtrmi th lowr liftim spcificatio limit. (ii) Udr symmtric tropy loss fuctio, calculat th aysia stimator

Mathmatics ttrs 08; 4(): 0-4 3 Whr T ˆ θ = ( β + T) ( + )( + ) = Xi. (iii) Calculat th 00 ( γ )% o-sid cofidc itrval [, ) for liftim prformac idx C, whr is dfid as quatio (). (iv) Th dcisio ruls of statistical tst ar providd as follows: If th prformac idx valu c [, ), w rjct to th ull hypothsis H0 ad coclud that th liftim prformac idx of product mts th rquird lvl. If th prformac idx valu c [, ),, w accpt th ull hypothsis H 0 ad coclud that th liftim prformac idx dos ot mt th rquird lvl. 5. Discussio To illustrat th fasibility ad practicability of th proposd aysia tstig mthod, a practical xampl proposd i rfrc Nlso [8] (98, P. 05. Tabl.) is adoptd. Thr is a lif tstig xprimt i which spcims of a typ of lctrical isulatig fluid wr subjct to a costat voltag strss. Th data st is rportd i Tabl. Tabl. if tstig xprimt data. Data St ( = 9 ) 0.9 0.78.3.78 0.96 4.5.06 6.50 3.75 3.6 4.85 7.89 3.5 4.67 7.35 8.7 8.0 33.9 36.7 alakrisha t al. [9] aalyzd ths =9 obsrvatios, ad thy provd that th data st has a xpotial distributio with pdf as () by usig last squars mthod ad goodss of fit tst. Gail ad Gastwirth [0] also provd th data st ca b modld by xpotial modl basd o Gii statistics. I th follow, w will giv th dtail stps of th proposd tstig procdur about C as: (i) Dtrmi th lowr liftim limit =.04. That is to say if th liftim of a lctrical isulatig fluid xcds hours, th th lctrical isulatig fluid is a coformig product. I ordr to satisfy th rgardig opratioal prformac. Th coformig rat P r is rquird to xcd 80 prct. Rfrrig to Tabl, th valu of C is rquird to xcd 0.80. Thus, th prformac idx targt valu is c=0.80. (ii) Establish tstig hypothsis as blow H : C 0.80 H : C > 0.80. 0 (iii) Spcifid a sigificac lvl γ = 0.05. (iv) Calculat th 00 (- γ )% o-sid cofidc itrval [, ) for C, whr, χ (( )) ( ˆ γ + = C) ( + ) Hr w suppos th prior distributio of θ is th oiformativ prior, i.. = β = 0. Th th 95% osidd cofidc itrval for C is [, ) = [0.8983, ). (v) caus of th prformac idx valu c = 0.80 [, ), w rjct th ull hypothsis H : 0.80 0 C. Thus, w ca coclud that th liftim prformac idx of products mts th rquird lvl. 6. Coclusios Procss capability idics ar widly mployd i maufactur idustry to assss th prformac ad pottial of thir procss. iftim prformac idx is o of th most importat PCIs. This papr studis th aysia stimatio ad aysia tst of lif prformac idx wh th liftim of products is distributd with xpotial distributio. Th proposd tstig mthod is asy to complt with th hlp of som softwar, such as xcl, matlab. Th w tst procdur ca provid rfrc for th trpris girs to assss whthr th tru prformac of products mts th rquirmts. Ackowldgmts This study is partially supportd by Social Scic Foudatio of Hua Provic (No. 5YA065). Th author is gratful to th rviwrs for a vry carful radig of th mauscript ad th suggstios. Rfrcs [] Shu M H, Wu H C. Maufacturig procss prformac valuatio for fuzzy data basd o loss-basd capability idx [J]. Soft Computig, 0, 6 ():89-99. [] Ch C C, ai C M, Ni H Y. Masurig procss capability idx Cpm with fuzzy data [J]. Quality & Quatity, 00, 44 (3):59-535. [3] Jura J M, Grya F M, igham R S J. Quality Cotrol Hadbook. Nw York: McGraw-Hill., 974. [4] Eslamipoor R, Hossii-Nasab H. A modifid procss capability idx usig loss fuctio cocpt [J]. Quality & Rliability Egirig Itratioal, 06, 3 ():435-44. [5] Ch K S, Huag C F, Chag T C. A mathmatical programmig modl for costructig th cofidc itrval of procss capability idx Cpm i valuatig procss prformac: a xampl of fiv-way pip [J]. Joural of th Chis Istitut of Egirs, 07, 40 ():6-33. [6] Gu X, Ma Y, iu J, t al. Robust paramtr dsig for multivariat quality charactristics basd o procss capability idx with idividual obsrvatios [J]. Systms Egirig & Elctroics, 07, 39 ():36-368.

4 Guobig Fa: aysia Tst for iftim Prformac Idx of Expotial Distributio udr Symmtric Etropy oss Fuctio [7] Abdolshah M, Yusuff R M, Hog T S, t al. Masurig procss capability idx Cpmk, with fuzzy data ad compar it with othr fuzzy procss capability idics [J]. Exprt Systms with Applicatios, 0, 38 (6):645-6457. [8] Kuriati N, Yh R H, Wu C W. A samplig schm for rsubmittd lots basd o o-sidd capability idics [J]. Quality Tchology & Quatitativ Maagmt, 05, (4):50-55. [9] Sifi S, Nzhad M S F. Variabl samplig pla for rsubmittd lots basd o procss capability idx ad aysia approach [J]. Th Itratioal Joural of Advacd Maufacturig Tchology, 07, 88 (9-): 547-555. [0] Motgomry D C. Itroductio to Statistical Quality Cotrol, Nw York: Joh Wily & Sos, 985. [] W C, Hog C W, Wu J W. Computatioal procdur of prformac assssmt of liftim idx of ormal products with fuzzy data udr th typ II right csord samplig pla [J]. Joural of Itlligt & Fuzzy Systms, 05, 8 (4):755-773. [] iu M. F., R H. P. aysia tst procdur of liftim prformac idx for xpotial distributio udr progrssiv typ-ii csorig. Itratioal Joural of Applid Mathmatics & Statistics, 03, 3 ():7-38. [3] Wu C W, Shu M H, Chag Y N. Variabl-samplig plas basd o liftim-prformac idx udr xpotial distributio with csorig ad its xtsios [J]. Applid Mathmatical Modllig, 08, 55:8-93. [4] Solima A E. Assssig th liftim prformac idx usig xpotiatd frcht distributio with th progrssiv firstfailur-csorig schm [J]. Amrica Joural of Thortical ad Applid Statistics, 04, 3 (6):67-76. [5] i. aysia tst for liftim prformac idx of Ailamujia distributio udr squard rror loss fuctio [J]. Pur ad Applid Mathmatics Joural, 06, 5 (6):8-85. [6] Zhao S, Sog Y, Sog, t al. Estimatio of ordrd mas of two sampl xpotial distributios udr symmtric tropy loss [J]. Joural of Jili Uivrsity, 007, 45 ():44-48. [7] i. ays stimatio of Topp-o distributio udr symmtric tropy loss fuctio basd o lowr rcord valus [J]. 06, 4 (6):84. [8] awlss J F. Statistical Modl ad Mthods for iftim Data. Joh Wily & Sos, Nw York, 98. [9] alakrisha N, i C T, ad Cha P S. A compariso of two simpl prdictio itrvals for xpotial distributio. IEEE Trasactios o Rliability, 005, 54: 7-33. [0] Gail M H ad Gastwirth J. A scal-fr goodss of fit tst for th xpotial distributio basd o th Gii statistics, Joural of th Royal Statistical Socity,, 978, 40: 350-357.