Software Project Effort Estimation Analysis

Abstract – The main aim of this research paper is to provide knowledge about software project effort estimation measures. Software project effort estimation plays very important role in the software development process. If the estimation is correct, the organization can complete its assignment within the predicted time. Otherwise, it will cause loss to the organization. So the quality and success of the software project depends on effort estimation. Also the fast changing nature of the software development, change of languages and the change of IDE can make the effort prediction failures. There are different types of estimation techniques. But we cannot select any one method as the best method for all kinds of projects. In this paper some of

There are estimation methods ranging from old classical models to current machine learning methods. These models are purely based on either linear regression or non-linear regression. Such models take only size of the project as input [9] [16]. One example of such model is COCOMO. COCOMO and SLIM models are also known as empirical models and these are the popular models [1][6]. Other than simple COCOMO model, remaining models take some additional parameters for estimating accurate effort. Though more models are incorporated with many parameters, complexity of the estimation process is also increased. Nowadays, to improve the accuracy of the estimation algorithmic models are combined with analogy based and machine learning based estimation processes. Here we have discussed three types of estimation techniques.

2. Categories of Estimation

Software project effort estimation models are used for project budgeting, planning, scheduling and risk analysis [17]. There are two major steps in determining how long a project will take and how much it will cost. The first is to estimate its size; the second is to use size along with other environmental factors to estimate effort and its associated cost. Sizing is the prediction of coding needed to fulfill requirements. Estimation is the prediction of resources needed to complete the project of predicted size, taking into account factors

It uses 7 to 17 cost drivers to estimate the effort. COCOMO II includes the Application Composition Model (for early prototyping efforts) and Early Design and Post- Architecture models.

The COCOMO II estimation is also based on the important parameter size which refers to the size of the software measured in SLOC or KSLOC.

Effort = 2.94 * EAF * (size) Exp (5)

Where

EAF is the Effort Adjustment Factor derived from the Cost Drivers.

Exp is an exponent derived from the five Scale Drivers.

The 5 Scale Drivers are:

• Precedence precedentness

• Development Flexibility

• Architecture / Risk Resolution

• Team Cohesion

• Process Maturity

4.3 SLIM

This is Putnam’s Software Life cycle Model which is an empirical estimation model [15]. This was proposed by Lawerence H.Putnam in 1978 in the field of software process modeling. The equation for Putnam’s model is: S = Env x (Effort)1/3td4/3 (6)

Where, Env is the environment indicator, td is the time of delivery, Effort is expressed in person-year and size S is expressed in line of