ESCOM/April 2001Aristotle University/Singular Int’l1 BRACE: BootstRap based Analogy Cost Estimation Automated support for an enhanced effort prediction.

Slides:



Advertisements
Similar presentations
Efficiency and Productivity Measurement: Bootstrapping DEA Scores
Advertisements

Review bootstrap and permutation
Introduction to Propensity Score Matching
Hypothesis testing and confidence intervals by resampling by J. Kárász.
Copyright 2000, Stephan Kelley1 Estimating User Interface Effort Using A Formal Method By Stephan Kelley 16 November 2000.
Autocorrelation and Linkage Cause Bias in Evaluation of Relational Learners David Jensen and Jennifer Neville.
The Comparison of the Software Cost Estimating Methods
1 Calibrating Function Points Using Neuro-Fuzzy Technique Vivian Xia NFA Estimation Inc. London, Ontario, Canada Danny Ho IT Department.
Prediction Basic concepts. Scope Prediction of:  Resources  Calendar time  Quality (or lack of quality)  Change impact  Process performance  Often.
Introduction to Statistics
University of Southern California Center for Systems and Software Engineering 1 © USC-CSSE A Constrained Regression Technique for COCOMO Calibration Presented.
2008 Chingchun 1 Bootstrap Chingchun Huang ( 黃敬群 ) Vision Lab, NCTU.
Stat 301 – Day 37 Bootstrapping, cont (5.5). Last Time - Bootstrapping A simulation tool for exploring the sampling distribution of a statistic, using.
STAT 572: Bootstrap Project Group Members: Cindy Bothwell Erik Barry Erhardt Nina Greenberg Casey Richardson Zachary Taylor.
Bootstrapping applied to t-tests
Bootstrap spatobotp ttaoospbr Hesterberger & Moore, chapter 16 1.
AutoSimOA : A Framework for Automated Analysis of Simulation Output Stewart Robinson Katy Hoad, Ruth Davies Funded by.
Scot Exec Course Nov/Dec 04 Ambitious title? Confidence intervals, design effects and significance tests for surveys. How to calculate sample numbers when.
Getting Business Value out of (Test) Metrics Richard Terry (UK) Rob Baarda (NL)
S Neuendorf 2004 Prediction of Software Defects SASQAG March 2004 by Steve Neuendorf.
2015 AprilUNIVERSITY OF HAIFA, DEPARTMENT OF STATISTICS, SEMINAR FOR M.A 1 Hastie, Tibshirani and Friedman.The Elements of Statistical Learning (2nd edition,
Cmpe 589 Spring Software Quality Metrics Product  product attributes –Size, complexity, design features, performance, quality level Process  Used.
1 Validation & Verification Chapter VALIDATION & VERIFICATION Very Difficult Very Important Conceptually distinct, but performed simultaneously.
The Effects of Ranging Noise on Multihop Localization: An Empirical Study from UC Berkeley Abon.
Chap 20-1 Statistics for Business and Economics, 6e © 2007 Pearson Education, Inc. Chapter 20 Sampling: Additional Topics in Sampling Statistics for Business.
Bootstrapping – the neglected approach to uncertainty European Real Estate Society Conference Eindhoven, Nederlands, June 2011 Paul Kershaw University.
Lecture 12 Statistical Inference (Estimation) Point and Interval estimation By Aziza Munir.
Biostatistics IV An introduction to bootstrap. 2 Getting something from nothing? In Rudolph Erich Raspe's tale, Baron Munchausen had, in one of his many.
PARAMETRIC STATISTICAL INFERENCE
University of Ottawa - Bio 4118 – Applied Biostatistics © Antoine Morin and Scott Findlay 08/10/ :23 PM 1 Some basic statistical concepts, statistics.
Software Estimation How hard can it be? Peter R Hill.
Using Resampling Techniques to Measure the Effectiveness of Providers in Workers’ Compensation Insurance David Speights Senior Research Statistician HNC.
Experimental Evaluation of Learning Algorithms Part 1.
Lecture 4 Software Metrics
Matching Estimators Methods of Economic Investigation Lecture 11.
Resampling techniques
Sampling And Resampling Risk Analysis for Water Resources Planning and Management Institute for Water Resources May 2007.
Introduction to Software Project Estimation I (Condensed) Barry Schrag Software Engineering Consultant MCSD, MCAD, MCDBA Bellevue.
Limits to Statistical Theory Bootstrap analysis ESM April 2006.
Bootstraps and Jackknives Hal Whitehead BIOL4062/5062.
- 1 - Overall procedure of validation Calibration Validation Figure 12.4 Validation, calibration, and prediction (Oberkampf and Barone, 2004 ). Model accuracy.
WERST – Methodology Group
1 Experience from Studies of Software Maintenance and Evolution Parastoo Mohagheghi Post doc, NTNU-IDI SEVO Seminar, 16 March 2006.
Random Forests Ujjwol Subedi. Introduction What is Random Tree? ◦ Is a tree constructed randomly from a set of possible trees having K random features.
Timothy Aman, FCAS MAAA Managing Director, Guy Carpenter Miami Statistical Limitations of Catastrophe Models CAS Limited Attendance Seminar New York, NY.
Case Selection and Resampling Lucila Ohno-Machado HST951.
Sampling and Statistical Analysis for Decision Making A. A. Elimam College of Business San Francisco State University.
Selecting a Sample. outline Difference between sampling in quantitative & qualitative research.
(6) Estimating Computer’s efficiency Software Estimation The objective of Software Estimation is to provide the skills needed to accurately predict the.
Ex St 801 Statistical Methods Inference about a Single Population Mean (CI)
Project Plan Task 8 and VERSUS2 Installation problems Anatoly Myravyev and Anastasia Bundel, Hydrometcenter of Russia March 2010.
1/61: Topic 1.2 – Extensions of the Linear Regression Model Microeconometric Modeling William Greene Stern School of Business New York University New York.
Graduate School of Information Science, Nara Institute of Science and Technology - Wed. 7 April 2004Profes 2004 Effort Estimation Based on Collaborative.
Quantifying Uncertainty
Bootstrapping James G. Anderson, Ph.D. Purdue University.
Chapter 9 Estimation using a single sample. What is statistics? -is the science which deals with 1.Collection of data 2.Presentation of data 3.Analysis.
WELCOME TO BIOSTATISTICS! WELCOME TO BIOSTATISTICS! Course content.
Estimating standard error using bootstrap
Application of the Bootstrap Estimating a Population Mean
Kaniz Rashid Lubana Mamun MS Student: CSU Hayward Dr. Eric A. Suess
2. Industry 4.0: novel sensors, control algorithms, and servo-presses
Project Management for Software Engineers (Summer 2017)
Simulation: Sensitivity, Bootstrap, and Power
Quantifying uncertainty using the bootstrap
L. Isella, A. Karvounaraki (JRC) D. Karlis (AUEB)
Bootstrap Confidence Intervals using Percentiles
Methods of Economic Investigation Lecture 12
Phase Distribution of Software Development Effort
BOOTSTRAPPING: LEARNING FROM THE SAMPLE
Ch13 Empirical Methods.
Presentation transcript:

ESCOM/April 2001Aristotle University/Singular Int’l1 BRACE: BootstRap based Analogy Cost Estimation Automated support for an enhanced effort prediction method I. Stamelos, L. Angelis Aristotle Univ. Thessaloniki E. Sakellaris Singular International

ESCOM/April 2001Aristotle University/Singular Int’l2 Cost Estimation Methods Expert judgement (experience – based estimation) Algorithmic cost estimation (statistical models such as regression) Estimation by analogy (Case Based Reasoning-comparison)

ESCOM/April 2001Aristotle University/Singular Int’l3 Estimation by Analogy (EbA) Characterise new project by certain attributes and place it into a historical data set

ESCOM/April 2001Aristotle University/Singular Int’l4 Estimation by Analogy (cont’d) Calculate distances of the new project from the completed ones and find few "neighbour" projects. Estimate the unknown effort by a location statistic (mean, median) of the efforts of the neighbour projects

ESCOM/April 2001Aristotle University/Singular Int’l5 Distance Metrics

ESCOM/April 2001Aristotle University/Singular Int’l6 Efficiency of EbA Shepperd and Schofield (1997): Superiority of EbA when compared to OLS regression models with 9 industrial datasets (Angel Tool used for EbA calibration) Other researchers (Myrtveit-Stensrud, Briand et al., Jeffery et al.) have recently observed contradicting results

ESCOM/April 2001Aristotle University/Singular Int’l7 Research Issues in EbA Method Accuracy –Extend calibration options: choice of distance metric –Application of EbA with proper historical data Generation of Interval Estimates –Calculation of Confidence Intervals (%CI) –Other measures of accuracy (bias, etc) Angelis, Stamelos, ‘A Simulation Tool for...’, EMSE, March 2000

ESCOM/April 2001Aristotle University/Singular Int’l8 Bootstrap Non parametric bootstrap: –Draw with replacement from the sample, a large number of new samples (of same size) –Estimate each time the effort of the new project –Use the empirical distribution (or an estimation) of the bootstrap samples in order to obtain confidence intervals Parametric bootstrap: based on the multivariate distribution of the original dataset

ESCOM/April 2001Aristotle University/Singular Int’l9 EbA calibration with Bootstrap MMRE - PRED(25) Distributions (Albrecht data set)

ESCOM/April 2001Aristotle University/Singular Int’l10 Confidence Interval Estimation with Bootstrap

ESCOM/April 2001Aristotle University/Singular Int’l11 Comparison of EbA and Regression Confidence Intervals (Abran-Robilland Data Set)

ESCOM/April 2001Aristotle University/Singular Int’l12 BRACE Functions Definition of attributes and project characterization Project/attribute management (e.g. exclusion of projects/attributes from calculations) Calibration of EbA (with and without bootstrap) including the various distance metric options Generation of estimations for a single project (with and without bootstrap) Typical utility functions and file management facilities

ESCOM/April 2001Aristotle University/Singular Int’l13

ESCOM/April 2001Aristotle University/Singular Int’l14

ESCOM/April 2001Aristotle University/Singular Int’l15 A case-study on software projects for the industry The ISBSG Cost Data Base  International Software Benchmarking Standards Group (Australia)  Non profit organization collecting software project data from around the world  Release 6 contains 789 software projects from 20 countries

ESCOM/April 2001Aristotle University/Singular Int’l16 ISBSG Project Data Project Nature (Organisation Type, Business Area Type, Application Type, …) Project Work Effort Data (man-hours) Project Size Data (Function Points) Project Quality Data (defects)...

ESCOM/April 2001Aristotle University/Singular Int’l17 Supply Chain ISBSG Project Subset  59 projects implementing information systems for manufacturing, logistics, warehouse management, …  characterized through effort, size, elapsed time, team size, project nature attributes  accurate project attribute measurement  average productivity ~ 190 FP/ 1000 mh

ESCOM/April 2001Aristotle University/Singular Int’l18 BRACE Application  Various strategies were tried because of missing values in project characterisation  Best strategy pursued a trade-off between number of projects and attributes  Precision was measured through project jackniving  Different treatment for elapsed time and max team size

ESCOM/April 2001Aristotle University/Singular Int’l19 EbA Precision Results Best parameter configuration: 30 projects, Canberra distance, one analogy, size adjustment: MMRE = 28.84%, PRED(25) = 46.67% When using elapsed time and team size Minkowski, λ=3 distance MMRE = 23.84%, PRED(25) = 70.37%

ESCOM/April 2001Aristotle University/Singular Int’l20 Future work  Project portfolio estimation  Clustering of the cost dataset  Implementation of Parametric Bootstrap  Optimisation techniques in calibration  Replication of the study with new ISBSG release (~1000 projects)