Data Driven Defect Predication for Upcoming new Projects in Software Houses

Every single software development organization produces and uses measurement data in order to plan, design and develop software products e.g., software size in terms of function points/source lines of code, number of defects. The main objectives of using measurement data are: 1) characterization of

Project Title

Data Driven Defect Predication for Upcoming new Projects in Software Houses

Project Area of Specialization

Software Engineering

Project Summary

Every single software development organization produces and uses measurement data in order to plan, design and develop software products e.g., software size in terms of function points/source lines of code, number of defects. The main objectives of using measurement data are: 1) characterization of software processes, products and resources, 2) comparing actual state versus desired/planned state of software, 3) making improvements in software processes, products and resources based on the data, 4) making data-driven predictions of software attributes (e.g., size, effort, defect) in future products based on historical data [1–3]. In this project, our objective is to develop a tool that enables the data-driven predictions of software attributes for software project managers.

The proposed tool will use data mining and machine learning methods for prediction. According to Tom Mitchell machine learning is defined as, " A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E" [4]. In our tool, we will get experience (E) from available datasets of software organizations (e.g., ISBSG, NASA) by performing tasks (T) in terms of supervised and unsupervised learning, the standard evaluation measures (P) such as precision, recall, and F1-measure will be used for comparison results of prediction tasks (T) and we will improve prediction results by providing mechanism for pre-processing of dataset and statistical analysis of results to the project managers. The overall process of designing the tool will be done according to guidelines of CRISP methodology. [5]

Most software development companies implement measurement processes with help of top-down approach by collecting data to understand, evaluate and improve software development processes [6, 7]. However, using already collected data to predict future products is still a challenge. Our tool will help software organizations to use their own data for predicting the attributes (e.g., size, effort, defects) in future products. The idea of using their own tool is beneficial because of two reasons: 1) There are no standard metrics in software development industry which can be used to predict a specific attribute [8–10], 2) In addition, the prediction/estimation methods (e.g., COCOMO, IFPUG, and COSMIC) are developed based on the regression analysis in different software industries of USA and Europe. The adaptations of such methods in a different context have different challenges e.g., discarding few of the collected metrics or needing to collect new metrics because available regression-based prediction methods only take specific software metrics. Therefore, there is a need of tool which could help the companies to use their own collected data for prediction.

Project Objectives

Our main goal is to provide better planning, estimation and prediction of upcoming software projects in software organizations. It will predict attributes in future projects and also provide automated estimation process to make better planning.

Our objective is to use existing dataset efficiently so that better prediction could be made.  We will perform pre-processing techniques on dataset and statistical analysis on the result to provide reliable analysis of predictions.

Project Implementation Method

We Used Different tools/technologies list below:

• React JS for Frontend
• Node
• Django Framework
• MongoDB
• Machine Learning
• Data mining

Benefits of the Project

This tool will help mangers by providing different stats that are found from the data provided by the manger to the tool such as risk analysis of project, defect prediction, manpower prediction, software lifetime, framework suitable for the proposed project. It’ll allows manager to get stats with less to no knowledge of AI, machine learning, data mining etc. by providing him with a graphical user interface to perform all the functions. This tool will automate already implemented estimation methods such as COCOMO, IFPUG, and COSMIC etc.

Technical Details of Final Deliverable

The Final Deliverables includes a website from which users can use and get their work done

Final Deliverable of the Project

Software System

Core Industry

IT

Other Industries

Others

Core Technology

Others

Other Technologies

Sustainable Development Goals

Industry, Innovation and Infrastructure

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com