A probabilistic model for predicting software development effort

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

We use the naive Bayes model to forecast software effort A causal model is developed from the literature, and a procedure to learn Bayesian prior and conditional probabilities is provided. Using a data set of 40 real-life software projects we test our model. Our results indicate that the probabilistic forecasting models allow managers to estimate joint probability distribution over different software effort estimates. A software project manager may use the joint probability distribution to develop a cumulative probability distribution, which in turn may help the manager estimate the uncertainty that the project effort may be greater than the estimated effort.

Original languageEnglish (US)
Pages (from-to)581-588
Number of pages8
JournalLecture Notes in Computer Science
Volume2668
StatePublished - 2003

Fingerprint

Statistical Models
Probabilistic Model
Software Development
Software engineering
Software
Probability distributions
Managers
Probability Distribution
Joint Distribution
Estimate
Causal Model
Prior Probability
Naive Bayes
Conditional probability
Uncertainty
Forecast
Forecasting
Model

All Science Journal Classification (ASJC) codes

  • Computer Science(all)
  • Biochemistry, Genetics and Molecular Biology(all)
  • Theoretical Computer Science

Cite this

@article{abc8080b621742b480d7eca380f7a574,
title = "A probabilistic model for predicting software development effort",
abstract = "We use the naive Bayes model to forecast software effort A causal model is developed from the literature, and a procedure to learn Bayesian prior and conditional probabilities is provided. Using a data set of 40 real-life software projects we test our model. Our results indicate that the probabilistic forecasting models allow managers to estimate joint probability distribution over different software effort estimates. A software project manager may use the joint probability distribution to develop a cumulative probability distribution, which in turn may help the manager estimate the uncertainty that the project effort may be greater than the estimated effort.",
author = "Pendharkar, {Parag C.} and Girish Subramanian and Rodger, {James A.}",
year = "2003",
language = "English (US)",
volume = "2668",
pages = "581--588",
journal = "Lecture Notes in Computer Science",
issn = "0302-9743",
publisher = "Springer Verlag",

}

A probabilistic model for predicting software development effort. / Pendharkar, Parag C.; Subramanian, Girish; Rodger, James A.

In: Lecture Notes in Computer Science, Vol. 2668, 2003, p. 581-588.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A probabilistic model for predicting software development effort

AU - Pendharkar, Parag C.

AU - Subramanian, Girish

AU - Rodger, James A.

PY - 2003

Y1 - 2003

N2 - We use the naive Bayes model to forecast software effort A causal model is developed from the literature, and a procedure to learn Bayesian prior and conditional probabilities is provided. Using a data set of 40 real-life software projects we test our model. Our results indicate that the probabilistic forecasting models allow managers to estimate joint probability distribution over different software effort estimates. A software project manager may use the joint probability distribution to develop a cumulative probability distribution, which in turn may help the manager estimate the uncertainty that the project effort may be greater than the estimated effort.

AB - We use the naive Bayes model to forecast software effort A causal model is developed from the literature, and a procedure to learn Bayesian prior and conditional probabilities is provided. Using a data set of 40 real-life software projects we test our model. Our results indicate that the probabilistic forecasting models allow managers to estimate joint probability distribution over different software effort estimates. A software project manager may use the joint probability distribution to develop a cumulative probability distribution, which in turn may help the manager estimate the uncertainty that the project effort may be greater than the estimated effort.

UR - http://www.scopus.com/inward/record.url?scp=33646494541&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33646494541&partnerID=8YFLogxK

M3 - Article

VL - 2668

SP - 581

EP - 588

JO - Lecture Notes in Computer Science

JF - Lecture Notes in Computer Science

SN - 0302-9743

ER -