Statistics for HEP Roger Barlow Manchester University

Slides:



Advertisements
Similar presentations
Statistics for HEP Roger Barlow Manchester University Lecture 4: Confidence Intervals.
Advertisements

DO’S AND DONT’S WITH LIKELIHOODS Louis Lyons Oxford (CDF)
Unified approach to the classical statistical analysis of small signals Gary J. Feldman and Robert D. Cousins.
27 th March CERN Higgs searches: CL s W. J. Murray RAL.
1 Methods of Experimental Particle Physics Alexei Safonov Lecture #21.
1 LIMITS Why limits? Methods for upper limits Desirable properties Dealing with systematics Feldman-Cousins Recommendations.
An Answer and a Question Limits: Combining 2 results Significance: Does  2 give  2 ? Roger Barlow BIRS meeting July 2006.
Setting Limits in the Presence of Nuisance Parameters Wolfgang A Rolke Angel M López Jan Conrad, CERN.
1 Do’s and Dont’s with L ikelihoods Louis Lyons Oxford and IC CDF IC, February 2008.
Statistics In HEP 2 Helge VossHadron Collider Physics Summer School June 8-17, 2011― Statistics in HEP 1 How do we understand/interpret our measurements.
8. Statistical tests 8.1 Hypotheses K. Desch – Statistical methods of data analysis SS10 Frequent problem: Decision making based on statistical information.
Statistics for Particle Physics: Intervals Roger Barlow Karlsruhe: 12 October 2009.
Results from Small Numbers Roger Barlow Manchester 26 th September 2007.
1 Do’s and Dont’s with L ikelihoods Louis Lyons Oxford CDF CERN, October 2006.
Basics of Statistical Estimation. Learning Probabilities: Classical Approach Simplest case: Flipping a thumbtack tails heads True probability  is unknown.
1 Nuisance parameters and systematic uncertainties Glen Cowan Royal Holloway, University of London IoP Half.
Credibility of confidence intervals Dean Karlen / Carleton University Advanced Statistical Techniques in Particle Physics Durham, March 2002.
G. Cowan Lectures on Statistical Data Analysis Lecture 12 page 1 Statistical Data Analysis: Lecture 12 1Probability, Bayes’ theorem 2Random variables and.
1 Do’s and Dont’s with L ikelihoods Louis Lyons Oxford CDF Manchester, 16 th November 2005.
Statistics for Particle Physics: Limits Roger Barlow Karlsruhe: 12 October 2009.
G. Cowan 2011 CERN Summer Student Lectures on Statistics / Lecture 41 Introduction to Statistics − Day 4 Lecture 1 Probability Random variables, probability.
Inference about a Mean Part II
G. Cowan Lectures on Statistical Data Analysis Lecture 10 page 1 Statistical Data Analysis: Lecture 10 1Probability, Bayes’ theorem 2Random variables and.
G. Cowan Lectures on Statistical Data Analysis 1 Statistical Data Analysis: Lecture 7 1Probability, Bayes’ theorem, random variables, pdfs 2Functions of.
G. Cowan Discovery and limits / DESY, 4-7 October 2011 / Lecture 3 1 Statistical Methods for Discovery and Limits Lecture 3: Limits for Poisson mean: Bayesian.
Everything You Always Wanted To Know About Limits* Roger Barlow Manchester University YETI06 *But were afraid to ask.
Statistical Analysis of Systematic Errors and Small Signals Reinhard Schwienhorst University of Minnesota 10/26/99.
880.P20 Winter 2006 Richard Kass 1 Confidence Intervals and Upper Limits Confidence intervals (CI) are related to confidence limits (CL). To calculate.
G. Cowan 2009 CERN Summer Student Lectures on Statistics1 Introduction to Statistics − Day 4 Lecture 1 Probability Random variables, probability densities,
Statistics In HEP Helge VossHadron Collider Physics Summer School June 8-17, 2011― Statistics in HEP 1 How do we understand/interpret our measurements.
Lecture 4: Statistics Review II Date: 9/5/02  Hypothesis tests: power  Estimation: likelihood, moment estimation, least square  Statistical properties.
Confidence Intervals First ICFA Instrumentation School/Workshop At Morelia, Mexico, November 18-29, 2002 Harrison B. Prosper Florida State University.
Probability, Statistics and Errors in High Energy Physics Wen-Chen Chang Institute of Physics, Academia Sinica 章文箴 中央研究院 物理研究所.
1 Methods of Experimental Particle Physics Alexei Safonov Lecture #25.
1 Methods of Experimental Particle Physics Alexei Safonov Lecture #24.
G. Cowan Lectures on Statistical Data Analysis Lecture 8 page 1 Statistical Data Analysis: Lecture 8 1Probability, Bayes’ theorem 2Random variables and.
Statistical Inference Statistical inference is concerned with the use of sample data to make inferences about unknown population parameters. For example,
1 Introduction to Statistics − Day 3 Glen Cowan Lecture 1 Probability Random variables, probability densities, etc. Brief catalogue of probability densities.
G. Cowan Lectures on Statistical Data Analysis Lecture 4 page 1 Lecture 4 1 Probability (90 min.) Definition, Bayes’ theorem, probability densities and.
G. Cowan Computing and Statistical Data Analysis / Stat 9 1 Computing and Statistical Data Analysis Stat 9: Parameter Estimation, Limits London Postgraduate.
2005 Unbinned Point Source Analysis Update Jim Braun IceCube Fall 2006 Collaboration Meeting.
In Bayesian theory, a test statistics can be defined by taking the ratio of the Bayes factors for the two hypotheses: The ratio measures the probability.
G. Cowan Lectures on Statistical Data Analysis Lecture 12 page 1 Statistical Data Analysis: Lecture 12 1Probability, Bayes’ theorem 2Random variables and.
Ivo van Vulpen Why does RooFit put asymmetric errors on data points ? 10 slides on a ‘too easy’ topic that I hope confuse, but do not irritate you
Tom.h.wilson Department of Geology and Geography West Virginia University Morgantown, WV.
Confidence Limits and Intervals 3: Various other topics Roger Barlow SLUO Lectures on Statistics August 2006.
Frequentist Confidence Limits and Intervals Roger Barlow SLUO Lectures on Statistics August 2006.
Statistics (2) Fitting Roger Barlow Manchester University IDPASC school Sesimbra 14 th December 2010.
Lecture Slides Elementary Statistics Twelfth Edition
arXiv:physics/ v3 [physics.data-an]
χ2 and Goodness of Fit & Likelihood for Parameters
BAYES and FREQUENTISM: The Return of an Old Controversy
Confidence Intervals and Limits
Likelihoods 1) Introduction 2) Do’s & Dont’s
Statistics for the LHC Lecture 3: Setting limits
Hypothesis Testing Is It Significant?.
Lecture 4 1 Probability (90 min.)
CHAPTER 21: Comparing Two Means
Incorporating systematic uncertainties into upper limits
Where did we stop? The Bayes decision rule guarantees an optimal classification… … But it requires the knowledge of P(ci|x) (or p(x|ci) and P(ci)) We.
Lecture 4 1 Probability Definition, Bayes’ theorem, probability densities and their properties, catalogue of pdfs, Monte Carlo 2 Statistical tests general.
Statistical Tests and Limits Lecture 2: Limits
Computing and Statistical Data Analysis / Stat 7
Introduction to Unfolding
Introduction to Statistics − Day 4
Lecture 4 1 Probability Definition, Bayes’ theorem, probability densities and their properties, catalogue of pdfs, Monte Carlo 2 Statistical tests general.
Bayesian Statistics on a Shoestring Assaf Oron, May 2008
Mathematical Foundations of BME Reza Shadmehr
Everything You Always Wanted To Know About Limits*
Computing and Statistical Data Analysis / Stat 10
Presentation transcript:

Statistics for HEP Roger Barlow Manchester University Lecture 4: Confidence Intervals

The Straightforward Example Apples of different weights Need to describe the distribution  = 68g  = 17 g All weights between 24 and 167 g (Tolerance) 90% lie between 50 and 100 g 94% are less than 100 g 96% are more than 50 g Confidence level statements 50 100

Confidence Levels Can quote at any level (68%, 95%, 99%…) Upper or lower or twosided (x<U x<L L<x<U) Two-sided has further choice (central, shortest…) U

The Frequentist Twist Particles of the same weight Distribution spread by measurement errors What can we say about M?  = 68  = 17 “M<90” or “M>55” or “60<M<85” @90% CL These are each always true or always false Solution: Refer to ensemble of statements 50 100

Frequentist CL in detail You have a meter: no bias, Gaussian error 0.1. For a value XT it gives a value XM according to a Gaussian Distribution XM is within 0.1 of XT 68% of the time XT is within 0.1 of XM 68% of the time Can state XM-0.1<XT<XM+0.1 @68% CL

Confidence Belts Construct Horizontally Read Vertically For more complicated distributions it isn’t quite so easy But the principle is the same True X Measured X

Coverage “L<x<U” @ 95% confidence (or “x>L” or “x<U”) This statement belongs to an ensemble of similar statements of which at least* 95% are true 95% is the coverage This is a statement about U and L, not about x. *Maybe more. Overcoverage EG composite hypotheses. Meter with resolution  0.1 Think about: the difference between a 90% upper limit and the upper limit of a 90% central interval.

Discrete Distributions May be unable to select (say) 5% region Play safe. Gives overcoverage CL belt edges become steps Binomial: see tables True continuous p or l Poisson Given 2 events, if the true mean is 6.3 (or more) then the chance of getting a fluctuation this low (or lower) is only 5% (or less) Measured discrete N

Problems for Frequentists Poisson: Signal + Background Background mean 2.50 Detect 3 events: Total < 6.68 @ 95% Signal<4.18@95% Detect 0 events Total < 2.30 @ 95% Signal < -0.20 @ 95% Weigh object+container with some Gaussian precision Get reading R R- < M+C < R+ @68% R-C- < M < R-C+ @68% E.g. C=50, R=141, =10 81<M<101 @68% E.g. C=50, R=55, =10 -5 < M < 5 @68% E.g. C=50, R=31, =10 -29 < M < -9 @68% These statements are OK. We are allowed to get 32% / 5% wrong. But they are stupid

Gives same limits as Frequentist method for simple Gaussian Bayes to the rescue Standard (Gaussian) measurement No prior knowledge of true value No prior knowledge of measurement result P(Data|Theory) is Gaussian P(Theory|Data) is Gaussian Interpret this with Probability statements in any way you please xtrue Gives same limits as Frequentist method for simple Gaussian

Bayesian Confidence Intervals (contd) Observe (say) 2 events P(;2)P(2; )=e- 2 Normalise and interpret If you know background mean is 1.7, then you know >1.7 Multiply, normalise and interpret  2

Bayes: words of caution Taking prior in  as flat is not justified Can argue for prior flat in ln  or 1/  or whatever Good practice to try a couple of priors to see if it matters

Feldman-Cousins Unified Method Physicists are human Ideal Physicist Choose Strategy Examine data Quote result Real Physicist Quote Result Example: You have a background of 3.2 Observe 5 events? Quote one-sided upper limit (9.27-3.2 =6.07@90%) Observe 25 events? Quote two-sided limits

“Flip-Flopping” 1 sided 2 sided S N S N S Allowed S Allowed N S This is not a true confidence belt! Coverage varies. N

Solution: Construct belt that does the flip-flopping For 90% CL For every S select set of N-values in belt Total probability must sum to 90% (or more): there are many strategies for doing this S N Crow & Gardner strategy (almost right): Select N-values with highest probability  shortest interval

Better Strategy To construct band for a given S: For all N: N is Poisson from S+B B known, may be large E.g. B=9.2,S=0 and N=1 P=.1% - not in C-G band But any S>0 will be worse To construct band for a given S: For all N: Find P(N;S+B) and Pbest=P(N;N) if (N>B) else P(N;B) Rank on P/Pbest Accept N into band until  P(N;S+B) 90% Fair comparison of P is with best P for this N Either at S=N-B or S=0

Feldman and Cousins Summary Makes us more honest (a bit) Avoids forbidden regions in a Frequentist way Not easy to calculate Has to be done separately for each value of B Can lead to 2-tailed limits where you don’t want to claim a discovery Weird effects for N=0; larger B gives lower (=better) upper limit

Maximum Likelihood and Confidence Levels ML estimator (large N) has variance given by MVB At peak For large N Ln L is a parabola (L is a Gaussian) Ln L Falls by ½ at a Falls by 2 at Read off 68% , 95% confidence regions

MVB example N Gaussian measurements: estimate  Ln L given by Differentiate twice wrt  Take expectation value – but it’s a constant Invert and negate:

Another MVB example N Gaussian measurements: estimate  Ln L still given by Differentiate twice wrt  Take expectation value <(xi-)2>=  2  i Gives Invert and negate:

ML for small N ln L is not a parabola a’ a So just do it directly Argue: we could (invariance) transform to some a’ for which it is a parabola We could/should then get limits on a’ using standard Lmax-½ technique These would translate to limits on a These limits would be at the values of a for which L= Lmax-½ So just do it directly

Multidimensional ML L is multidimensional Gaussian For 2-d 39.3% lies within 1 i.e. within region bounded by L=Lmax-½ For 68% need L=Lmax-1.15 Construct region(s) to taste using numbers from integrated 2 distribution b a

Confidence Intervals Descriptive Frequentist Bayesian Feldman-Cousins technique Bayesian Maximum Likelihood Standard Asymmetric Multidimensional