Estimating software effort and function point using regression, Support Vector Machine and Artificial Neural Networks models

Accurate computation of software effort, cost and time required ahead would greatly reduce risk and maximize profit. Estimating software effort or computing the required function point helps project manager to better estimate the time and budget required for a project. Many statistical models were proposed in the past. These models suffer many problems related to parameter estimation and structure determination of the models. In this paper we presents two models for software effort estimation and one model for function points using Linear Regression (LR), Support Vector Machines (SVM) and Artificial Neural Networks (ANN). The proposed models have number of inputs and single output. The first model utilizes the Source Line Of Code (KLOC) as inputs; while the second model utilize the KLOC and development Methodology (ME) as inputs to estimate the Effort (E); while the third model utilize the Inputs, Outputs, Files, and User Inquiries as inputs to estimate the Function Point (FP). The proposed SVM and ANN models show better estimation capabilities compared to linear regression model models. These models are capable of providing better assistant to software project manager in computing the effort required of the number of function points.