Electronic Thesis/Dissertation
 

Using Predictive Analytics to Deliver an Improved IT Project Cost Performance Model

Open Access

This research presents the application of various single machine learning algorithms (generalized linear models, random forest, decision tree, gradient boosted trees, support vector machine, deep Learning) and ensemble (bagging) to predict the cost and schedule of a software project. The software project data used was obtained from the ISBSG 2019 R1, which has 9,186 project instances (rows) and 253 attributes (columns). The data preprocessing activities (data reduction, feature selection, data transformation, and missing data imputation) were conducted to prepare the dataset as a relevant input data for the ML application. For predicting the cost, 624 project instances with 17 active attributes (predictors) were identified using the backward elimination feature selection (BWE-FS). The input data for schedule prediction consists of 621 project instances, with 16 active predictors identified based on the BWE-FS. All the models were developed using 10-fold cross-validation, which was implemented in RapidMiner studio version 9.7. The numerical prediction performance of the models was evaluated using the Root Mean square error (RMSE), Mean Absolute Error (MAE), Relative Error (RE), Root Relative Squared Error (RRSE), Correlation coefficient (r), Squared Correlation (R2), and PRED(25). The decision tree accuracy showed that 91.35% of the predicted cost and 90.18% of the predicted schedule were within 25% of the actual values. Ensemble bagged trees showed that 91.67% of the predicted cost and 94.20% of the predicted schedule were within 25% of the actual values. This suggests that the ensemble approach consistently improved the single MLA's prediction accuracy for the software project cost and schedule.

Author Language Date created Type of Work License
  • All rights reserved
Rights statement GW Unit Degree Advisor Committee Member(s) Persistent URL

Notice to Authors

If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.

Thumbnail Title Date Uploaded Visibility Actions
Preview of Awolumate_gwu_0075A_15359.pdf Awolumate_gwu_0075A_15359.pdf 2020-12-03 Open Access