WG 3: Verification and applications of ensemble forecasts
|
|
- Erik Moody
- 5 years ago
- Views:
Transcription
1 WG 3: Verificatin and applicatins f ensemble frecasts Participants: Ken Mylne (Chair), Martin Leutbecher (secretary), Magdalena A. Balmaseda, Pac Dblas-Reyes, Lizzie Frude, Jse A. Garcia-Mya, Anna Ghelli, Renate Hagedrn, Edit Hagel, Flrian Pappenberger, Fernand Prates, Anders Perssn, Cristina Prim, Kamal Puri, Thmas Schumann, Olivier Talagrand, Jutta Thielen, Helen Titley First, the discussins f the wrking grup n the Verificatin f Ensemble Frecasts and the Applicatin f Ensemble frecasts are summarised in Sectins 1 and 2, respectively. This is fllwed by specific recmmendatins fr ECMWF in Sectin 3. Mst f these recmmendatins are als cnsidered t be relevant fr ther prducers f ensemble frecasts. 1. Verificatin f Ensemble Frecasts 1.1. Purpse f Verificatin The wrking grup identified three different purpses f the verificatin: Objective measure fr imprving the EPS and guide future develpments f system and funding Guide frecasters, service prviders and users n which prduct(s) t trust and use. Hwever, sme participants thught that, at present, nt many frecasters lk at verificatin scres fr prbabilistic frecasts. Demnstrate usefulness f ensemble predictin systems fr particular applicatins r varius user grups (frm general public t decisin makers) Statistical Significance f Results It was recmmended that cnfidence intervals shuld be prvided in all verificatin statistics. Further research will be required in rder t study methds f cmputing cnfidence intervals and significance tests and t verify the reliability f cnfidence intervals. Examples: Btstrap methds, analytic methds. Latter require assumptins abut the distributin. Btstrap requires fewer assumptins but may be cstly t calculate Accunting fr the uncertainty f the verificatin data There was cnsensus that it is necessary t accunt fr bservatin errrs/analysis errrs in the prbabilistic verificatin. One accepted way f ding this (fr rank-histgram and prbabilistic verificatin in general) is t add nise with the same statistics as that f the verifying data t the individual ensemble frecasts (Saetra et al). A secnd alternative is t use a decnvlutin methd (Bwler). A third methd is t cnsider the bservatin as a prbability distributin (Talagrand & Candille). The first methd will tell us nly hw well we can predict an bservatin nt hw well we can predict the true state. It was nted that it can be difficult t estimate the errr characteristics f sme verificatin data: e.g. rain gauge data Chice f verificatin measures There are many different verificatin measures in use, and the grup did nt attempt t review them all. Discussin centred arund hw t cmmunicate perfrmance effectively t decisin-makers and users wh may nt be specialist in the details f the science. Wrkshp n Ensemble Predictin, 7-9 Nvember
2 The chice f the verificatin will reflect its purpse (see abve). There was general cnsensus that classic upper air verificatin has its value in prviding guidance cncerning the general accuracy and reliability f a (prbabilistic) predictin system. It shuld be cmplemented by verificatin f surface weather variables which are particularly relevant t users and fr shrt-range ensemble predictin. The wrking grup agreed that several measures are required t examine different aspects f the Prbability Density Functin (PDF). Fr instance, the rank histgram, n the ne hand, and the reliability cmpnent f the Brier scre with respect t a threshld, n the ther hand, measure different aspects f reliability. A single summary measure (althugh desirable frm a management perspective) was cnsidered inadequate. It was nted that reliability is a statistical prperty, and glbal reliability, as estimated by a particular measure ver a given set f realizatins f an EPS, may always result frm mutual cmpensatin between individually unreliable subsets (Example: flatness f a rank-histgram is a necessary but nt a sufficient cnditin fr the statistical reliability f an EPS). The decmpsitin f scres t better understand results was encuraged where apprpriate fr the audience (e.g. reliability and reslutin cmpnent f Brier-type scres) Cncerning recmmendatins fr system upgrades, selective use f particular scres was discuraged as this may encurage the selectin f the favurable subsets. In ther wrds, the impact f system upgrades (e-suite versus -suite) shuld always be dcumented by the same standard set f scres (further discussin is required which nes these shuld be). Use f Reliability Diagrams (including sharpness f the a psteriri calibrated prbabilities) was cnsidered t be the easiest way t cmmunicate prbabilistic verificatin t nn-specialist audiences. There is need fr care in the use f the Relative Operating Characteristic, and the area under the ROC in particular, as a verificatin measure. It is particularly difficult t explain t users and nnspecialists. Value f ROC area is very sensitive t hw it is calculated and t small changes in perfrmance. Use f cnfidence measures wuld help. Prbabilistic scres need t be cmplemented by mre general mdel validatin: (ability t simulate climatlgical mean and variance, ) 1.5. Avidance f False Skill (added by Chair) It was nted, but nt discussed in detail due t time cnstraints, that withut due care verificatin can indicate mre skill in frecast systems than is warranted due t climatlgical differences between lcatins included in the verificatin set - the base rate effect. This was illustrated by Tm Hamill s presentatin in the wrkshp. It was nted that ECMWF has already taken steps t avid this in sme f the verificatin presented at the wrkshp. Tw methds are prpsed t minimise the effect: Define events fr prbabilistic verificatin as percentiles f the climatlgical distributin rather than abslute values [eg. p(t>90 th percentile) r p(t> 1 s.d. abve nrmal) but nt p(t>5 Celsius) ]. (This methd is prpsed by Hamill and Juras, and was used by ECMWF in sme verificatin presented at the wrkshp.) Prpsed by Hamill in his presentatin: grup sites accrding t their climatlgical frequency f the event (eg grup all sites fr which 5mm/24h ccurs with climatlgical frequencies f 0-5%, 6-15%, 16-25%, etc). Estimate the gegraphical area fr which each climatlgical categry is 2 Wrkshp n Ensemble Predictin, 7-9 Nvember 2007
3 representative. Calculate verificatin statistics fr each grup f sites, and then average results weighting the values accrding t the prprtin f gegraphical area represented by each grup Verificatin f rare/extreme events Sme discussin fcussed n whether a different methdlgy needs t be adpted fr the verificatin f rare/extreme events (t be distinguished frm high-impact events which need nt be rare in the climatlgical sense) There is a prblem f sample size fr rare events; n statistical significance may be reached. The same is true fr events which ccur almst always because f the symmetry f scres (event ccurring / nt ccurring). The predictin f events within a finite space-time dmain selected arund a particular weather event r atmspheric feature (as ppsed t pint values) will lead t higher prbabilities fr sme extreme events than lcal prbabilities (example cyclne strike prbabilities). Sme extraplatin may be pssible frm the verificatin fr the mre mderate threshlds which will have statistically mre reliable results. Use f errr bars wuld help guide hw far it is reasnable t extraplate cnclusins. Case studies were cnsidered t be essential in assessing perfrmance fr extreme events. It was prpsed that case studies shuld include cases where ensembles predict nn-zer prbabilities f extreme events, but which d nt verify, t ensure a reliable prbabilistic predictin. This shuld be used t cmplement bjective statistical verificatin f less extreme events. It was nted that where extreme events d ccur, the bservatin is very ften clse t the extreme f the ensemble distributin - hence users shuld be strngly encuraged t pay attentin t lw prbability alerts. This has als implicatins fr decisin making (cst/lss analysis). Supprt frm high-reslutin mdels may strengthen the signal. Fr sme cases f severe events, experience suggests that the actual predictability may be higher than implied by the EPS. Since predictability is a prperty f a frecast system it is f interest t quantify hw the ensemble dispersin relates t the frecast accuracy f state-f-the-art deterministic mdels Aspects relevant fr the verificatin/applicatins f seasnal/mnthly frecasts The questin was raised whether an effrt shuld be made t unify verificatin tls used fr different applicatins (e.g. medium-range, mnthly, seasnal predictins). The limitatins f btaining statistically significant verificatin results fr seasnal and lnger predictins was discussed A member f the wrking grup nted that the repetitin f similar anmalies in subsequent years in the seasnal frecast may decrease trust f users in the useful signal in this prduct. It was suggested that the repetitin f similar anmalies might be due t the chice f climate and its lack f accunting fr the climate change trend. It was suggested that the climatlgy shuld include a trend; this aspect is relevant bth fr the verificatin and the cmmunicatin f seasnal frecasts.
4 Case studies: It was recmmended t identify a priri events where the predicted PDF deviates significantly frm the climatlgical PDF t avid a selective verificatin f nly the events that did verify (see discussin abve n extreme events). Discussin was held arund whether seasnal frecasts shuld be issued in areas r seasns with little r n skill. It is always imprtant t cmmunicate infrmatin n the level f skill. In cases f n skill it is preferable that n frecast shuld be issued, but the user culd be prvided with the climatlgy. Where there is sme skill and frecasts are issued, it was recmmended that frecasts shuld be issued cnsistently, including thse ccasins when there is n strng signal. In this case the frecast shuld revert t climatlgy tgether with the infrmatin that there is n strng r useful signal in the seasnal frecast. In summary, issue frecasts where there is skill but n signal, but NOT where there is signal but n skill. It was mentined that the cnsistency f subsequent frecasts can increase trust f users in the prduct (as an example the mnthly frecast was mentined). It was briefly discussed whether the cmmunicatin f seasnal predictins in the frm f anmalies with respect t the recent N years (with N being sme number up t 10) may be mre useful and/r easier t interpret fr sme users than anmalies with respect t a lnger term climate which includes perids beynd the persnal memry f the users. This was cnsidered particularly useful where the climate has undergne significant change in recent decades Benchmark(s) Sme fair cmparisn f EPS with prbability distributins built frm the High-Reslutin deterministic frecast shuld be examined. A simple example f such prbability distributins are Gaussian distributins centred n the deterministic frecast with a standard deviatin depending n frecast range Educatinal Aspects f Verificatin It was felt that mre educatin/explanatin f the meaning f scres and changes f the scres was required. A shrt guide t the meaning/usefulness f different verificatin measures and their apprpriateness fr different applicatins wuld be useful. When results f EPS verificatin are presented, a range f scres is required. The relative meaning f different scres must be explained as the apparent meaning f sme results may be cunterintuitive (a well knwn example fr deterministic frecasts is the reductin f RMS errr when the activity f the mdel is reduced and vice versa). 2. Applicatins f Ensemble Frecasts 2.1. Cmmunicatin f prbabilistic frecasts A WMO guide n the cmmunicatin f prbabilistic frecasts will be published sn. (It is nt published n web yet please cntact ken.mylne@metffice.gv.uk fr update) Use and limitatin f Ensemble Mean. There was sme disagreement n the value f the ensemble mean. Sme experience suggested that it was useful in intrducing use f ensembles int predminantly deterministic envirnments. Hwever, there was cnsiderable cncern abut the limitatins f the mean, in particular its inability ever t predict extreme events and the view was 4 Wrkshp n Ensemble Predictin, 7-9 Nvember 2007
5 als expressed that the ensemble mean shuld never be used. Where used it shuld always be accmpanied by prbabilities f extreme r high-impact events Hydrlgical applicatins The need fr re-frecasts was stressed nt nly fr calibratin but als fr verificatin (the frmer shuld be cvered by WG2) A limited sample size is an issue fr fld frecasting. Fr instance, the definitin f a climatlgical PDF fr streamflw frm a catchment pses a prblem. Thus, it is difficult t evaluate skill scres r bjectively cmpare different ensemble cnfiguratins There may be particular verificatin difficulties arising frm the effects f river cntrl measures and changing river prfiles which impact the cnsistency f the bservatins Recmmendatins cncerning design and testing f frecast system Based n predictability thery, medium-range and later range frecasts shuld be issued in prbabilistic frm. Therefre, the prime aim f ECMWF shuld be t prvide the tls t predict a reliable and sharp PDF f the atmspheric state rather than just a single deterministic high-reslutin frecast. The PDF shuld be based n all available infrmatin, i.e. the EPS and the high-reslutin deterministic frecast. Cnsequently, the EPS shuld be given the same level f attentin as the deterministic frecast system. Research n hw t best cmbine high-reslutin deterministic frecast and EPS shuld be cntinued at ECMWF. It was, hwever, als stated that a user/applicatin-specific cmbinatin may be superir t a generic cmbinatin. In such cases, the prductin f an ptimally cmbined PDF wuld fall int the respnsibility f member states r individual end-users. The EPS shuld therefre be affrded: Sufficient cmputer resurce t allw ptimal mdel perfrmance tuning f the frecast mdel/physical parameterisatins duratin f experimental suites It was felt that it shuld be further investigated whether there is a benefit f ging frm 62 vertical levels t 91 in the EPS (eg benefit f imprved stratspheric reslutin n medium-range frecast). 3. Summary f Recmmendatins fr ECMWF and ther EPS Prducers 3.1. Recmmendatins n Verificatin Cnfidence intervals shuld be prvided in all verificatin statistics. Sme research is required in the best techniques fr estimating cnfidence intervals, but there is already sme useful wrk in the literature. Methds fr accunting fr bservatin r analysis errr shuld be applied in calculatin f verificatin results. Several methds are prpsed in the literature. Methds shuld be used t minimise the impact f apparent false skill in verificatin results caused by differences in climatlgical frequency f events (base rate) at different lcatins. Several measures are required t examine different aspects f EPS perfrmance. A single summary measure, althugh desirable frm a management perspective, is inadequate.
6 These several measures shuld be used in a cnsistent fashin. The impact f prpsed system changes shuld be dcumented using the full set t give a balanced picture f the strengths and weaknesses f the change; selective use f particular scres is strngly discuraged. Reliability diagrams, tgether with sharpness diagrams f the crrespnding a psteriri calibrated prbabilities, prvide an effective way t cmmunicate prbabilistic perfrmance t nn-specialists. The Relative Operating Characteristic, while useful in research, shuld be used with care as it is difficult t explain and the ROC area summary measure can be very sensitive t hw it is calculated and is frequently mis-interpreted. Statistically significant verificatin f rare extreme events is impssible. Judicius use f cnfidence intervals may allw sme extraplatin f results frm less extreme events. Use f case studies fr extreme events shuld be balanced with cases where ensembles indicated a prbability f a severe event but nne verified. Fair cmparisn shuld be made between EPS frecasts and the high-reslutin deterministic frecast dressed with errr statistics. Sme investigatin is encuraged f whether the ensemble spread reflects the true predictability f extreme events frm the high-reslutin deterministic mdel. A simple summary guide f cmmnly used prbabilistic verificatin statistics and their interpretatin is required Recmmendatins n Applicatins The prime aim f ECMWF (r ther NWP systems) shuld be t prvide tls t predict a reliable and sharp prbability distributin f the future state f the atmsphere based n all available infrmatin the needs f EPS shuld therefre be affrded equal attentin as high-reslutin deterministic frecasts, and apprpriate levels f resurces. Seasnal frecast shuld be issued cnsistently, and where there is n signal this shuld be clearly cmmunicated. One shuld cnsider the pssibility f describing the trend in the reference climatlgy fr seasnal frecasts, t avid issuing frecasts which are dminated by the climate change trend. 6 Wrkshp n Ensemble Predictin, 7-9 Nvember 2007
Distributions, spatial statistics and a Bayesian perspective
Distributins, spatial statistics and a Bayesian perspective Dug Nychka Natinal Center fr Atmspheric Research Distributins and densities Cnditinal distributins and Bayes Thm Bivariate nrmal Spatial statistics
More informationChecking the resolved resonance region in EXFOR database
Checking the reslved resnance regin in EXFOR database Gttfried Bertn Sciété de Calcul Mathématique (SCM) Oscar Cabells OECD/NEA Data Bank JEFF Meetings - Sessin JEFF Experiments Nvember 0-4, 017 Bulgne-Billancurt,
More informationModelling of Clock Behaviour. Don Percival. Applied Physics Laboratory University of Washington Seattle, Washington, USA
Mdelling f Clck Behaviur Dn Percival Applied Physics Labratry University f Washingtn Seattle, Washingtn, USA verheads and paper fr talk available at http://faculty.washingtn.edu/dbp/talks.html 1 Overview
More informationMath Foundations 20 Work Plan
Math Fundatins 20 Wrk Plan Units / Tpics 20.8 Demnstrate understanding f systems f linear inequalities in tw variables. Time Frame December 1-3 weeks 6-10 Majr Learning Indicatrs Identify situatins relevant
More informationENSC Discrete Time Systems. Project Outline. Semester
ENSC 49 - iscrete Time Systems Prject Outline Semester 006-1. Objectives The gal f the prject is t design a channel fading simulatr. Upn successful cmpletin f the prject, yu will reinfrce yur understanding
More informationPerfrmance f Sensitizing Rules n Shewhart Cntrl Charts with Autcrrelated Data Key Wrds: Autregressive, Mving Average, Runs Tests, Shewhart Cntrl Chart
Perfrmance f Sensitizing Rules n Shewhart Cntrl Charts with Autcrrelated Data Sandy D. Balkin Dennis K. J. Lin y Pennsylvania State University, University Park, PA 16802 Sandy Balkin is a graduate student
More informationBootstrap Method > # Purpose: understand how bootstrap method works > obs=c(11.96, 5.03, 67.40, 16.07, 31.50, 7.73, 11.10, 22.38) > n=length(obs) >
Btstrap Methd > # Purpse: understand hw btstrap methd wrks > bs=c(11.96, 5.03, 67.40, 16.07, 31.50, 7.73, 11.10, 22.38) > n=length(bs) > mean(bs) [1] 21.64625 > # estimate f lambda > lambda = 1/mean(bs);
More informationDocument for ENES5 meeting
HARMONISATION OF EXPOSURE SCENARIO SHORT TITLES Dcument fr ENES5 meeting Paper jintly prepared by ECHA Cefic DUCC ESCOM ES Shrt Titles Grup 13 Nvember 2013 OBJECTIVES FOR ENES5 The bjective f this dcument
More informationCAUSAL INFERENCE. Technical Track Session I. Phillippe Leite. The World Bank
CAUSAL INFERENCE Technical Track Sessin I Phillippe Leite The Wrld Bank These slides were develped by Christel Vermeersch and mdified by Phillippe Leite fr the purpse f this wrkshp Plicy questins are causal
More information, which yields. where z1. and z2
The Gaussian r Nrmal PDF, Page 1 The Gaussian r Nrmal Prbability Density Functin Authr: Jhn M Cimbala, Penn State University Latest revisin: 11 September 13 The Gaussian r Nrmal Prbability Density Functin
More informationSkill forecasting from different wind power ensemble prediction methods
Skill frecasting frm different wind pwer ensemble predictin methds Pierre Pinsn 1, Henrik Aa. Nielsen 1, Henrik Madsen 1 and Gerge Karinitakis 2 1 Technical University f Denmark, Infrmatics and Mathematical
More informationk-nearest Neighbor How to choose k Average of k points more reliable when: Large k: noise in attributes +o o noise in class labels
Mtivating Example Memry-Based Learning Instance-Based Learning K-earest eighbr Inductive Assumptin Similar inputs map t similar utputs If nt true => learning is impssible If true => learning reduces t
More informationHypothesis Tests for One Population Mean
Hypthesis Tests fr One Ppulatin Mean Chapter 9 Ala Abdelbaki Objective Objective: T estimate the value f ne ppulatin mean Inferential statistics using statistics in rder t estimate parameters We will be
More informationChapter 3: Cluster Analysis
Chapter 3: Cluster Analysis } 3.1 Basic Cncepts f Clustering 3.1.1 Cluster Analysis 3.1. Clustering Categries } 3. Partitining Methds 3..1 The principle 3.. K-Means Methd 3..3 K-Medids Methd 3..4 CLARA
More informationBASD HIGH SCHOOL FORMAL LAB REPORT
BASD HIGH SCHOOL FORMAL LAB REPORT *WARNING: After an explanatin f what t include in each sectin, there is an example f hw the sectin might lk using a sample experiment Keep in mind, the sample lab used
More informationWe respond to each of ORR s specific consultation questions in Annex A to this letter.
Je Quill Office f Rail Regulatin One Kemble Street Lndn, WC2B 4AN Hannah Devesn Regulatry Refrm Specialist Netwrk Rail Kings Place, 90 Yrk Way Lndn, N1 9AG Email:hannah.devesn@netwrkrail.c.uk Telephne:
More informationResampling Methods. Chapter 5. Chapter 5 1 / 52
Resampling Methds Chapter 5 Chapter 5 1 / 52 1 51 Validatin set apprach 2 52 Crss validatin 3 53 Btstrap Chapter 5 2 / 52 Abut Resampling An imprtant statistical tl Pretending the data as ppulatin and
More informationCHAPTER 24: INFERENCE IN REGRESSION. Chapter 24: Make inferences about the population from which the sample data came.
MATH 1342 Ch. 24 April 25 and 27, 2013 Page 1 f 5 CHAPTER 24: INFERENCE IN REGRESSION Chapters 4 and 5: Relatinships between tw quantitative variables. Be able t Make a graph (scatterplt) Summarize the
More informationCHAPTER 4 DIAGNOSTICS FOR INFLUENTIAL OBSERVATIONS
CHAPTER 4 DIAGNOSTICS FOR INFLUENTIAL OBSERVATIONS 1 Influential bservatins are bservatins whse presence in the data can have a distrting effect n the parameter estimates and pssibly the entire analysis,
More informationSUPPLEMENTARY MATERIAL GaGa: a simple and flexible hierarchical model for microarray data analysis
SUPPLEMENTARY MATERIAL GaGa: a simple and flexible hierarchical mdel fr micrarray data analysis David Rssell Department f Bistatistics M.D. Andersn Cancer Center, Hustn, TX 77030, USA rsselldavid@gmail.cm
More informationWe say that y is a linear function of x if. Chapter 13: The Correlation Coefficient and the Regression Line
Chapter 13: The Crrelatin Cefficient and the Regressin Line We begin with a sme useful facts abut straight lines. Recall the x, y crdinate system, as pictured belw. 3 2 1 y = 2.5 y = 0.5x 3 2 1 1 2 3 1
More informationWRITING THE REPORT. Organizing the report. Title Page. Table of Contents
WRITING THE REPORT Organizing the reprt Mst reprts shuld be rganized in the fllwing manner. Smetime there is a valid reasn t include extra chapters in within the bdy f the reprt. 1. Title page 2. Executive
More informationOn Huntsberger Type Shrinkage Estimator for the Mean of Normal Distribution ABSTRACT INTRODUCTION
Malaysian Jurnal f Mathematical Sciences 4(): 7-4 () On Huntsberger Type Shrinkage Estimatr fr the Mean f Nrmal Distributin Department f Mathematical and Physical Sciences, University f Nizwa, Sultanate
More informationNUROP CONGRESS PAPER CHINESE PINYIN TO CHINESE CHARACTER CONVERSION
NUROP Chinese Pinyin T Chinese Character Cnversin NUROP CONGRESS PAPER CHINESE PINYIN TO CHINESE CHARACTER CONVERSION CHIA LI SHI 1 AND LUA KIM TENG 2 Schl f Cmputing, Natinal University f Singapre 3 Science
More informationBOUNDED UNCERTAINTY AND CLIMATE CHANGE ECONOMICS. Christopher Costello, Andrew Solow, Michael Neubert, and Stephen Polasky
BOUNDED UNCERTAINTY AND CLIMATE CHANGE ECONOMICS Christpher Cstell, Andrew Slw, Michael Neubert, and Stephen Plasky Intrductin The central questin in the ecnmic analysis f climate change plicy cncerns
More informationAssessment Primer: Writing Instructional Objectives
Assessment Primer: Writing Instructinal Objectives (Based n Preparing Instructinal Objectives by Mager 1962 and Preparing Instructinal Objectives: A critical tl in the develpment f effective instructin
More informationA Quick Overview of the. Framework for K 12 Science Education
A Quick Overview f the NGSS EQuIP MODULE 1 Framewrk fr K 12 Science Educatin Mdule 1: A Quick Overview f the Framewrk fr K 12 Science Educatin This mdule prvides a brief backgrund n the Framewrk fr K-12
More informationThe blessing of dimensionality for kernel methods
fr kernel methds Building classifiers in high dimensinal space Pierre Dupnt Pierre.Dupnt@ucluvain.be Classifiers define decisin surfaces in sme feature space where the data is either initially represented
More informationTHE LIFE OF AN OBJECT IT SYSTEMS
THE LIFE OF AN OBJECT IT SYSTEMS Persns, bjects, r cncepts frm the real wrld, which we mdel as bjects in the IT system, have "lives". Actually, they have tw lives; the riginal in the real wrld has a life,
More informationA New Evaluation Measure. J. Joiner and L. Werner. The problems of evaluation and the needed criteria of evaluation
III-l III. A New Evaluatin Measure J. Jiner and L. Werner Abstract The prblems f evaluatin and the needed criteria f evaluatin measures in the SMART system f infrmatin retrieval are reviewed and discussed.
More informationEvaluating enterprise support: state of the art and future challenges. Dirk Czarnitzki KU Leuven, Belgium, and ZEW Mannheim, Germany
Evaluating enterprise supprt: state f the art and future challenges Dirk Czarnitzki KU Leuven, Belgium, and ZEW Mannheim, Germany Intrductin During the last decade, mircecnmetric ecnmetric cunterfactual
More informationComputational modeling techniques
Cmputatinal mdeling techniques Lecture 4: Mdel checing fr ODE mdels In Petre Department f IT, Åb Aademi http://www.users.ab.fi/ipetre/cmpmd/ Cntent Stichimetric matrix Calculating the mass cnservatin relatins
More informationAP Statistics Notes Unit Two: The Normal Distributions
AP Statistics Ntes Unit Tw: The Nrmal Distributins Syllabus Objectives: 1.5 The student will summarize distributins f data measuring the psitin using quartiles, percentiles, and standardized scres (z-scres).
More information5.4 Measurement Sampling Rates for Daily Maximum and Minimum Temperatures
5.4 Measurement Sampling Rates fr Daily Maximum and Minimum Temperatures 1 1 2 X. Lin, K. G. Hubbard, and C. B. Baker University f Nebraska, Lincln, Nebraska 2 Natinal Climatic Data Center 1 1. INTRODUCTION
More informationBiplots in Practice MICHAEL GREENACRE. Professor of Statistics at the Pompeu Fabra University. Chapter 13 Offprint
Biplts in Practice MICHAEL GREENACRE Prfessr f Statistics at the Pmpeu Fabra University Chapter 13 Offprint CASE STUDY BIOMEDICINE Cmparing Cancer Types Accrding t Gene Epressin Arrays First published:
More informationLand Information New Zealand Topographic Strategy DRAFT (for discussion)
Land Infrmatin New Zealand Tpgraphic Strategy DRAFT (fr discussin) Natinal Tpgraphic Office Intrductin The Land Infrmatin New Zealand Tpgraphic Strategy will prvide directin fr the cllectin and maintenance
More informationTechnical Bulletin. Generation Interconnection Procedures. Revisions to Cluster 4, Phase 1 Study Methodology
Technical Bulletin Generatin Intercnnectin Prcedures Revisins t Cluster 4, Phase 1 Study Methdlgy Release Date: Octber 20, 2011 (Finalizatin f the Draft Technical Bulletin released n September 19, 2011)
More informationTHERMAL-VACUUM VERSUS THERMAL- ATMOSPHERIC TESTS OF ELECTRONIC ASSEMBLIES
PREFERRED RELIABILITY PAGE 1 OF 5 PRACTICES PRACTICE NO. PT-TE-1409 THERMAL-VACUUM VERSUS THERMAL- ATMOSPHERIC Practice: Perfrm all thermal envirnmental tests n electrnic spaceflight hardware in a flight-like
More informationKinetic Model Completeness
5.68J/10.652J Spring 2003 Lecture Ntes Tuesday April 15, 2003 Kinetic Mdel Cmpleteness We say a chemical kinetic mdel is cmplete fr a particular reactin cnditin when it cntains all the species and reactins
More informationWhat is Statistical Learning?
What is Statistical Learning? Sales 5 10 15 20 25 Sales 5 10 15 20 25 Sales 5 10 15 20 25 0 50 100 200 300 TV 0 10 20 30 40 50 Radi 0 20 40 60 80 100 Newspaper Shwn are Sales vs TV, Radi and Newspaper,
More informationCS 477/677 Analysis of Algorithms Fall 2007 Dr. George Bebis Course Project Due Date: 11/29/2007
CS 477/677 Analysis f Algrithms Fall 2007 Dr. Gerge Bebis Curse Prject Due Date: 11/29/2007 Part1: Cmparisn f Srting Algrithms (70% f the prject grade) The bjective f the first part f the assignment is
More informationPattern Recognition 2014 Support Vector Machines
Pattern Recgnitin 2014 Supprt Vectr Machines Ad Feelders Universiteit Utrecht Ad Feelders ( Universiteit Utrecht ) Pattern Recgnitin 1 / 55 Overview 1 Separable Case 2 Kernel Functins 3 Allwing Errrs (Sft
More informationPart 3 Introduction to statistical classification techniques
Part 3 Intrductin t statistical classificatin techniques Machine Learning, Part 3, March 07 Fabi Rli Preamble ØIn Part we have seen that if we knw: Psterir prbabilities P(ω i / ) Or the equivalent terms
More informationWe can see from the graph above that the intersection is, i.e., [ ).
MTH 111 Cllege Algebra Lecture Ntes July 2, 2014 Functin Arithmetic: With nt t much difficulty, we ntice that inputs f functins are numbers, and utputs f functins are numbers. S whatever we can d with
More informationThe Kullback-Leibler Kernel as a Framework for Discriminant and Localized Representations for Visual Recognition
The Kullback-Leibler Kernel as a Framewrk fr Discriminant and Lcalized Representatins fr Visual Recgnitin Nun Vascncels Purdy H Pedr Mren ECE Department University f Califrnia, San Dieg HP Labs Cambridge
More informationAMERICAN PETROLEUM INSTITUTE API RP 581 RISK BASED INSPECTION BASE RESOURCE DOCUMENT BALLOT COVER PAGE
Ballt ID: Title: USING LIFE EXTENSION FACTOR (LEF) TO INCREASE BUNDLE INSPECTION INTERVAL Purpse: 1. Prvides a methd t increase a bundle s inspectin interval t accunt fr LEF. 2. Clarifies Table 8.6.5 Als
More informationo o IMPORTANT REMINDERS Reports will be graded largely on their ability to clearly communicate results and important conclusions.
BASD High Schl Frmal Lab Reprt GENERAL INFORMATION 12 pt Times New Rman fnt Duble-spaced, if required by yur teacher 1 inch margins n all sides (tp, bttm, left, and right) Always write in third persn (avid
More informationNWC SAF ENTERING A NEW PHASE
NWC SAF ENTERING A NEW PHASE Pilar Fernández Agencia Estatal de Meterlgía (AEMET) Lenard Priet Castr, 8; 28040 Madrid, Spain Phne: +34 915 819 654, Fax: +34 915 819 767 E-mail: mafernandeza@aemet.es Abstract
More informationTime, Synchronization, and Wireless Sensor Networks
Time, Synchrnizatin, and Wireless Sensr Netwrks Part II Ted Herman University f Iwa Ted Herman/March 2005 1 Presentatin: Part II metrics and techniques single-hp beacns reginal time znes ruting-structure
More informationMATCHING TECHNIQUES. Technical Track Session VI. Emanuela Galasso. The World Bank
MATCHING TECHNIQUES Technical Track Sessin VI Emanuela Galass The Wrld Bank These slides were develped by Christel Vermeersch and mdified by Emanuela Galass fr the purpse f this wrkshp When can we use
More informationT Algorithmic methods for data mining. Slide set 6: dimensionality reduction
T-61.5060 Algrithmic methds fr data mining Slide set 6: dimensinality reductin reading assignment LRU bk: 11.1 11.3 PCA tutrial in mycurses (ptinal) ptinal: An Elementary Prf f a Therem f Jhnsn and Lindenstrauss,
More informationMATHEMATICS SYLLABUS SECONDARY 5th YEAR
Eurpean Schls Office f the Secretary-General Pedaggical Develpment Unit Ref. : 011-01-D-8-en- Orig. : EN MATHEMATICS SYLLABUS SECONDARY 5th YEAR 6 perid/week curse APPROVED BY THE JOINT TEACHING COMMITTEE
More informationStatistics, Numerical Models and Ensembles
Statistics, Numerical Mdels and Ensembles Duglas Nychka, Reinhard Furrer,, Dan Cley Claudia Tebaldi, Linda Mearns, Jerry Meehl and Richard Smith (UNC). Spatial predictin and data assimilatin Precipitatin
More informationA Matrix Representation of Panel Data
web Extensin 6 Appendix 6.A A Matrix Representatin f Panel Data Panel data mdels cme in tw brad varieties, distinct intercept DGPs and errr cmpnent DGPs. his appendix presents matrix algebra representatins
More informationApplication of ILIUM to the estimation of the T eff [Fe/H] pair from BP/RP
Applicatin f ILIUM t the estimatin f the T eff [Fe/H] pair frm BP/RP prepared by: apprved by: reference: issue: 1 revisin: 1 date: 2009-02-10 status: Issued Cryn A.L. Bailer-Jnes Max Planck Institute fr
More informationSubject description processes
Subject representatin 6.1.2. Subject descriptin prcesses Overview Fur majr prcesses r areas f practice fr representing subjects are classificatin, subject catalging, indexing, and abstracting. The prcesses
More informationResults of an intercomparison of models of snowmelt runoff
Mdelling Snwmelt-Induced Prcesses (Prceedings f the Budapest Sympsium, July 1986). IAHS Publ. n. 155,1986. Results f an intercmparisn f mdels f snwmelt runff WRLD METERLGICAL RGANIZATIN CP N.5, 1211 Geneva
More informationPhysics 2B Chapter 23 Notes - Faraday s Law & Inductors Spring 2018
Michael Faraday lived in the Lndn area frm 1791 t 1867. He was 29 years ld when Hand Oersted, in 1820, accidentally discvered that electric current creates magnetic field. Thrugh empirical bservatin and
More informationSequential Allocation with Minimal Switching
In Cmputing Science and Statistics 28 (1996), pp. 567 572 Sequential Allcatin with Minimal Switching Quentin F. Stut 1 Janis Hardwick 1 EECS Dept., University f Michigan Statistics Dept., Purdue University
More informationUN Committee of Experts on Environmental Accounting New York, June Peter Cosier Wentworth Group of Concerned Scientists.
UN Cmmittee f Experts n Envirnmental Accunting New Yrk, June 2011 Peter Csier Wentwrth Grup f Cncerned Scientists Speaking Ntes Peter Csier: Directr f the Wentwrth Grup Cncerned Scientists based in Sydney,
More informationParticle Size Distributions from SANS Data Using the Maximum Entropy Method. By J. A. POTTON, G. J. DANIELL AND B. D. RAINFORD
3 J. Appl. Cryst. (1988). 21,3-8 Particle Size Distributins frm SANS Data Using the Maximum Entrpy Methd By J. A. PTTN, G. J. DANIELL AND B. D. RAINFRD Physics Department, The University, Suthamptn S9
More information22.54 Neutron Interactions and Applications (Spring 2004) Chapter 11 (3/11/04) Neutron Diffusion
.54 Neutrn Interactins and Applicatins (Spring 004) Chapter (3//04) Neutrn Diffusin References -- J. R. Lamarsh, Intrductin t Nuclear Reactr Thery (Addisn-Wesley, Reading, 966) T study neutrn diffusin
More informationWriting Guidelines. (Updated: November 25, 2009) Forwards
Writing Guidelines (Updated: Nvember 25, 2009) Frwards I have fund in my review f the manuscripts frm ur students and research assciates, as well as thse submitted t varius jurnals by thers that the majr
More informationWagon Markings Guidelines
Versin / Status: V 3.0 / apprved Wagn Markings Guidelines 1. Intrductin Article 4, para 4 f the Safety Directive (2004/49/EG amended by 2008/110/EC) stipulates the respnsibility f each manufacturer, maintenance
More informationMath Foundations 10 Work Plan
Math Fundatins 10 Wrk Plan Units / Tpics 10.1 Demnstrate understanding f factrs f whle numbers by: Prime factrs Greatest Cmmn Factrs (GCF) Least Cmmn Multiple (LCM) Principal square rt Cube rt Time Frame
More informationComprehensive Exam Guidelines Department of Chemical and Biomolecular Engineering, Ohio University
Cmprehensive Exam Guidelines Department f Chemical and Bimlecular Engineering, Ohi University Purpse In the Cmprehensive Exam, the student prepares an ral and a written research prpsal. The Cmprehensive
More informationProduct authorisation in case of in situ generation
Prduct authrisatin in case f in situ generatin Intrductin At the 74 th CA meeting (27-29 September 2017), Aqua Eurpa and ECA Cnsrtium presented their cncerns and prpsals n the management f the prduct authrisatin
More informationLab 1 The Scientific Method
INTRODUCTION The fllwing labratry exercise is designed t give yu, the student, an pprtunity t explre unknwn systems, r universes, and hypthesize pssible rules which may gvern the behavir within them. Scientific
More informationTHERMAL TEST LEVELS & DURATIONS
PREFERRED RELIABILITY PAGE 1 OF 7 PRACTICES PRACTICE NO. PT-TE-144 Practice: 1 Perfrm thermal dwell test n prtflight hardware ver the temperature range f +75 C/-2 C (applied at the thermal cntrl/munting
More informationAP Statistics Practice Test Unit Three Exploring Relationships Between Variables. Name Period Date
AP Statistics Practice Test Unit Three Explring Relatinships Between Variables Name Perid Date True r False: 1. Crrelatin and regressin require explanatry and respnse variables. 1. 2. Every least squares
More informationLesson Plan. Recode: They will do a graphic organizer to sequence the steps of scientific method.
Lessn Plan Reach: Ask the students if they ever ppped a bag f micrwave ppcrn and nticed hw many kernels were unppped at the bttm f the bag which made yu wnder if ther brands pp better than the ne yu are
More informationAssociated Students Flacks Internship
Assciated Students Flacks Internship 2016-2017 Applicatin Persnal Infrmatin: Name: Address: Phne #: Years at UCSB: Cumulative GPA: E-mail: Majr(s)/Minr(s): Units Cmpleted: Tw persnal references (Different
More informationPSU GISPOPSCI June 2011 Ordinary Least Squares & Spatial Linear Regression in GeoDa
There are tw parts t this lab. The first is intended t demnstrate hw t request and interpret the spatial diagnstics f a standard OLS regressin mdel using GeDa. The diagnstics prvide infrmatin abut the
More informationarxiv:hep-ph/ v1 2 Jun 1995
WIS-95//May-PH The rati F n /F p frm the analysis f data using a new scaling variable S. A. Gurvitz arxiv:hep-ph/95063v1 Jun 1995 Department f Particle Physics, Weizmann Institute f Science, Rehvt 76100,
More informationRN52-STK2 Starter Kit
The RN52-STK2 Starter Kit has everything yu need t kick-start yur prject and is a great tl fr develpers: Prf f cncept prttypes as well as final cmmercial slutins. fr educatinal use: Natural sciences curses
More informationUsing Bayesian Model Averaging to Calibrate Forecast Ensembles 1
Using Bayesian Mdel Averaging t Calibrate Frecast Ensembles 1 Adrian E. Raftery, Fadua Balabdaui, Tilmann Gneiting and Michael Plakwski Department f Statistics, University f Washingtn, Seattle, Washingtn
More informationStatistics Statistical method Variables Value Score Type of Research Level of Measurement...
Lecture 1 Displaying data... 12 Statistics... 13 Statistical methd... 13 Variables... 13 Value... 15 Scre... 15 Type f Research... 15 Level f Measurement... 15 Numeric/Quantitative variables... 15 Ordinal/Rank-rder
More informationLecture 2: Supervised vs. unsupervised learning, bias-variance tradeoff
Lecture 2: Supervised vs. unsupervised learning, bias-variance tradeff Reading: Chapter 2 STATS 202: Data mining and analysis September 27, 2017 1 / 20 Supervised vs. unsupervised learning In unsupervised
More informationDetermining the Accuracy of Modal Parameter Estimation Methods
Determining the Accuracy f Mdal Parameter Estimatin Methds by Michael Lee Ph.D., P.E. & Mar Richardsn Ph.D. Structural Measurement Systems Milpitas, CA Abstract The mst cmmn type f mdal testing system
More informationAIP Logic Chapter 4 Notes
AIP Lgic Chapter 4 Ntes Sectin 4.1 Sectin 4.2 Sectin 4.3 Sectin 4.4 Sectin 4.5 Sectin 4.6 Sectin 4.7 4.1 The Cmpnents f Categrical Prpsitins There are fur types f categrical prpsitins. Prpsitin Letter
More informationPopulation Structure and Migration Patterns of Atlantic Cod (Gadus morhua) in West Greenland Waters Based on Tagging Experiments from 1946 to 1964
NAFO Sci. Cun. Studies, : - Ppulatin Structure and Migratin Patterns f Atlantic Cd (Gadus mrhua) in West Greenland Waters Based n Tagging Experiments frm t Hlger Hvgard Greenland Fisheries Research Institute
More informationSIZE BIAS IN LINE TRANSECT SAMPLING: A FIELD TEST. Mark C. Otto Statistics Research Division, Bureau of the Census Washington, D.C , U.S.A.
SIZE BIAS IN LINE TRANSECT SAMPLING: A FIELD TEST Mark C. Ott Statistics Research Divisin, Bureau f the Census Washingtn, D.C. 20233, U.S.A. and Kenneth H. Pllck Department f Statistics, Nrth Carlina State
More informationInternal vs. external validity. External validity. This section is based on Stock and Watson s Chapter 9.
Sectin 7 Mdel Assessment This sectin is based n Stck and Watsn s Chapter 9. Internal vs. external validity Internal validity refers t whether the analysis is valid fr the ppulatin and sample being studied.
More informationLecture 2: Supervised vs. unsupervised learning, bias-variance tradeoff
Lecture 2: Supervised vs. unsupervised learning, bias-variance tradeff Reading: Chapter 2 STATS 202: Data mining and analysis September 27, 2017 1 / 20 Supervised vs. unsupervised learning In unsupervised
More informationCOMP 551 Applied Machine Learning Lecture 5: Generative models for linear classification
COMP 551 Applied Machine Learning Lecture 5: Generative mdels fr linear classificatin Instructr: Herke van Hf (herke.vanhf@mail.mcgill.ca) Slides mstly by: Jelle Pineau Class web page: www.cs.mcgill.ca/~hvanh2/cmp551
More informationDo big losses in judgmental adjustments affect experts behaviour? Fotios Petropoulos, Robert Fildes and Paul Goodwin
D big lsses in judgmental adjustments affect experts behaviur? Ftis Petrpuls, Rbert Fildes and Paul Gdwin This material has been created and cpyrighted by Lancaster Centre fr Frecasting, Lancaster University
More informationComparison of hybrid ensemble-4dvar with EnKF and 4DVar for regional-scale data assimilation
Cmparisn f hybrid ensemble-4dvar with EnKF and 4DVar fr reginal-scale data assimilatin Jn Pterjy and Fuqing Zhang Department f Meterlgy The Pennsylvania State University Wednesday 18 th December, 2013
More information3.4 Shrinkage Methods Prostate Cancer Data Example (Continued) Ridge Regression
3.3.4 Prstate Cancer Data Example (Cntinued) 3.4 Shrinkage Methds 61 Table 3.3 shws the cefficients frm a number f different selectin and shrinkage methds. They are best-subset selectin using an all-subsets
More informationChapter 1 Notes Using Geography Skills
Chapter 1 Ntes Using Gegraphy Skills Sectin 1: Thinking Like a Gegrapher Gegraphy is used t interpret the past, understand the present, and plan fr the future. Gegraphy is the study f the Earth. It is
More informationLead/Lag Compensator Frequency Domain Properties and Design Methods
Lectures 6 and 7 Lead/Lag Cmpensatr Frequency Dmain Prperties and Design Methds Definitin Cnsider the cmpensatr (ie cntrller Fr, it is called a lag cmpensatr s K Fr s, it is called a lead cmpensatr Ntatin
More informationCollege of Engineering Writing & Communication Resource Center
Cllege f Engineering Writing & Cmmunicatin Resurce Center 1250 BELLFLOWER BLVD. LONG BEACH, CA 90840 VIVIAN ENGINEERING CENTER 128B MS Thesis/Prject Wrkshp Handut The Scpe/Abstract The Abstract What? An
More informationComparison of two variable parameter Muskingum methods
Extreme Hydrlgical Events: Precipitatin, Flds and Drughts (Prceedings f the Ykhama Sympsium, July 1993). IAHS Publ. n. 213, 1993. 129 Cmparisn f tw variable parameter Muskingum methds M. PERUMAL Department
More informationHow do scientists measure trees? What is DBH?
Hw d scientists measure trees? What is DBH? Purpse Students develp an understanding f tree size and hw scientists measure trees. Students bserve and measure tree ckies and explre the relatinship between
More informationBios 6648: Design & conduct of clinical research
Bis 6648: Design & cnduct f clinical research Sectin 3 - Essential principle 3.1 Masking (blinding) 3.2 Treatment allcatin (randmizatin) 3.3 Study quality cntrl : Interim decisin and grup sequential :
More informationChemistry 20 Lesson 11 Electronegativity, Polarity and Shapes
Chemistry 20 Lessn 11 Electrnegativity, Plarity and Shapes In ur previus wrk we learned why atms frm cvalent bnds and hw t draw the resulting rganizatin f atms. In this lessn we will learn (a) hw the cmbinatin
More informationSnow avalanche runout from two Canadian mountain ranges
Annals f Glacilgy 18 1993 Internati n al Glaciigicai Sciety Snw avalanche runut frm tw Canadian muntain ranges D.J. NIXON AND D. M. MCCLUNG Department f Civil Engineering, University f British Clumbia,
More informationCOMP 551 Applied Machine Learning Lecture 9: Support Vector Machines (cont d)
COMP 551 Applied Machine Learning Lecture 9: Supprt Vectr Machines (cnt d) Instructr: Herke van Hf (herke.vanhf@mail.mcgill.ca) Slides mstly by: Class web page: www.cs.mcgill.ca/~hvanh2/cmp551 Unless therwise
More informationFive Whys How To Do It Better
Five Whys Definitin. As explained in the previus article, we define rt cause as simply the uncvering f hw the current prblem came int being. Fr a simple causal chain, it is the entire chain. Fr a cmplex
More informationApplications of latent trait theory to the development of norm-referenced tests.
University f Massachusetts Amherst SchlarWrks@UMass Amherst Dctral Dissertatins 1896 - February 2014 1-1-1979 Applicatins f latent trait thery t the develpment f nrm-referenced tests. Linda L. Ck University
More information