Machine Learning Model to Predict Automated Testing Adoption

1. Introduction

To sustain in this competitive era, IT companies are required to build long-lasting relationship with the customers and increase the reliability of their software (Qasim, 2017) which is only possible by producing high quality software. High quality software is one which satisfies customer expectations therefore it needs to be tested before delivering to the customer or client.

Software Testing is a task or an activity for checking the quality of the software or product which is under test. This activity helps finding the fault, defect or error in the software or system under test (SUT). It is an important and time consuming activity in a software development but due to its higher cost, it is often neglected. Companies who neglect and don’t focus on software testing can’t produce high quality software’s because it is an essential part in software development process. Approaches that can be used for testing are either Manual Testing or Automated Testing.

Manual Testing refers to the testing that is done by thoroughly going through the system and finding defects in it using critical thinking and developing scenarios. Manual Testing is time consuming and people/testers get fuzzy by repeating the same scenarios over and over again, hence the quality of software project gets compromised. However, Automated Testing refers to the testing that is done by using computer programs / scripts which are written for execution of tests. These scripts compare the actual outputs with the expected outputs and generate the results and reports that tell about the passing or failure of the test/feature/function or scenario. These tests run on the project/software and guides about the quality.

It is consider as myth that automated testing all the time results in improved software quality and should be used for every software development project (online, 2020) rather it depends on the many factors like project requirements nature and its team. Therefore we focused towards this area and developed machine learning model for prediction of automated testing adoption.

Many studies (Wang, 2018; Kasurinen & Smolander, 2010; Khari, 2020) presented the factors like project complexity, project time, project cost and team size which needs to be considered in order to use automated testing for certain software development project. These studies also helped us in our research for identifying factors. However, in this paper we categorize each factor in more detail and then presented machine learning model trained on data gathered about software development projects, its working team and intended users.

The main contributions of this paper are given below.

• •

  Categorization of factors which need to be consider for prediction of automated testing on certain software development projects.

• •

  Presented machine learning model for prediction of automated testing adoption.

The remaining paper sections are structured as follows: Section-2 of the paper explains literature. Section-3 provides the Methodology and Techniques that are used in the paper. Section-4 describes the results of the applied techniques and analysis of data then finally Section-5 provides conclusion of the paper and future work. Similarly, at the end, last section provided references.

2. Literature Review

Keeping the research objectives in mind, we selected relevant papers based on the following criteria:

• •

  Articles those are relevant to factors’ selection which effects or lead to automated testing maturity.

• •

  Articles that developed machine learning models using survey data.

• •

  We also included articles that describe how much test automation and techniques are effectively utilized in industry.