Novel Methods and Applications of Soft Computing in Software Effort Estimation

Abstract

ix newlineAbstract Software industries in general, product development enterprise in particular focus majorly on planning, scheduling and managing any and every software development project. The point of concern is the effort estimation. Proper estimation of the effort is very crucial particularly for long spanning software projects because accurate effort estimation will avoid any sort of possible failures. Here accurate estimation connotes neither high estimate nor low estimate because both of them are counterproductive. While very high estimates lead to the loss of competitiveness in the market, the low estimates may end up in a kind of disaster, as it lead to failure in meeting commitments, schedule delivery dates, and hence an economic loss. In this thesis, an explorative research done on developing predictive models for software effort estimation. The research work was mainly intended on conducting a unique active software development in a vaguely mimicked industry scenario. Three kinds of software projects collectively adequate in number were developed and the effort in terms of time consumption is actually measured. The said projects totalling around 150 were developed by post graduate students in three consecutive academic years 2015-16, 2016-17, and 2017-18. To have a flavor of different scenarios in project development, three kinds of projects namely visualization software projects, website development projects and technical and analytical based projects were developed. Among the three kinds of projects website development projects are full blown in terms of content, frontend and backend. The other two, may be considered as small scale projects owing to limitations of academic duration provided to the students. To the best of the knowledge of the researcher, this thesis reports an active research which is deemed to be unique and first of its kind. Following reasons attributes. and#61656; Most of the researchers (to an extent of 90%) have developed software effort estimation models using public domain datasets pertain