A Qualitative and Comprehensive Analysis of Software Testability Metrics and their Trends

Abstract

Advancement in technology has resulted in birth of critical and complex software. These require thorough testing to ensure production of reliable and high performance software. Testing is the most expensive part of the software life cycle and an estimate of testing efforts can result in smart utilization of resources. Testability is the ease of finding faults in a software and its estimate can reduce costs and increase life of the software. However the area lacks adequate research and standardisation. Current studies majorly report on Object Oriented paradigm and code level testability. This study aims to provide a broader review on Testability metrics, models and establishing relationship between program attributes and testability. Through this survey 29 studies have been selected for analysis. Our studies conclude that testability metrics at code level and design metric corresponding to size are commonly used. Relationships amongst these metrics with their test efforts are established using various machine learning models and presents testability trends with various program attributes. This comprehensive review helps in identifying suitable metric, expected trends with various program attributes and selection of suitable models to automate processes.