Software Cost Estimation

To study the process of Software Cost Estimation and its impact on the Software Development Process. We also highlight the various challenges involved in Software Cost Estimation and common solutions to navigate through these challenges. 

Background: Software Cost Estimation is widely considered to be a weak link in software project management. It requires a significant amount of effort to perform it correctly. Errors in Software Cost Estimation can be attributed to a variety of factors. Various studies in the last decade indicated that 3 out of 4 Software projects are not finished on time or within budget or both

Who is responsible for Software Cost Estimation? 
The group of people responsible for creating a software cost estimate can vary with each organization. However the following is possible in most scenarios -

- People who are directly involved with the implementation are involved in the 
..estimate.

- Project Manager is responsible for producing realistic cost estimates.

- Project Managers may perform this task on their own or consult with 
..programmers responsible.

- Various studies indicate that if the programmers responsible for development 
..are involved in the estimation it was more accurate. The programmers have 
..more motivation to meet the targets if they were involved in the estimation 
..process.

Following scenarios are also possible

- An independent cost estimation team creates an Estimate

- Independent Experts are given the Software specification and they create a 
..Software Cost estimate. The Estimation team reviews this and group 
..consensus arrives at a final figure.

Factors contributing to inaccurate estimation

· Scope Creeps, imprecise and drifting requirements 

· New software projects pose new challenges, which may be very different from 
..the past projects.

· Many teams fail to document metrics and lessons learned from past projects 
· Many a times the estimates are forced to match the available time and 
..resources by aggressive leaders 
· Unrealistic estimates may be created by various ‘political under currents’

Impact of Under-estimating: 

Under-Estimating a project can be vary damaging

- It leads to improper Project Planning

- It can also result in under-staffing and may result in an over worked and 
..burnt out team 
- Above all the quality of deliverables may be directly affected due insufficient 
..testing and QA 
- Missed Dead lines cause loss of Credibility and goodwill

The Estimation Process:
Generally the Software Cost estimation process comprises of 4 main steps:

1) Estimate the size of the development product.
This comprises of various sub-steps or sub tasks. These tasks may have been done already during Requirement Analysis phase. If not then they should be done as a part of the estimation Process. Important thing is that they should be done to ensure the success of the Estimation Process and the Software Project as a whole

a) Create a detailed Work Break Down Structure. This directly impacts the accuracy of the estimate. This is one of the most important steps. The Work Break down structure should include any and all tasks that are within the scope of the Project, which is being estimated. The most serious handicap is the inability to clearly visualize the steps involved in the Project. Executing a Software Project is not just coding.

b) The work Break down structure will include the size and complexity of each software module that can be expressed as number of Lines of Code, Function Points, or any other unit of measure

c) The Work Break down structure should include tasks other than coding such as Software Configuration Management, various levels and types of Testing, Documentation, Communication, User Interaction, Implementation, Knowledge Transition, Support tasks(if any) and so on

d) Clearly indicate or eliminate any gray areas (vague/unclear specifications etc.)

e) Also take into account the various Risk Factors and down times. There are many different Risk Factors involved – Technical aspects such as availability of the Environment, Server/Machine uptime, 3rd party Software Hardware failures or Human aspects – Employee Attrition, Sick time, etc. Some of them may seem to be ‘overkill’ but real world experience shows that these factors affect the time lines of a project. If ignored they may adversely impact the Project timelines and estimates.

2) Estimate the effort in person-hours. The Result of various tasks involved in step 1 is an effort estimate in person hours. The effort of various Project tasks expressed in person-hours is also influenced by various factors such as:

a) Experience/Capability of the Team members 
b) Technical resources 
c) Familiarity with the Development Tools and Technology Platform

3) Estimate the schedule in calendar months The Project Planners work closely with the Technical Leads, Project Manager and other stakeholders and create a Project schedule. Tight Schedules may impact the Cost needed to develop the Application.

4) Estimate the project cost in dollars (or other currency) 
Based on the above information the project effort is expressed in dollars or any other currency.

Measuring the Size/Complexity of the Software Program: 
This is one of the most elusive aspects in the Software Cost Estimation Process.
There are different methodologies for arriving at and expressing the size/complexity of the Software Program. Some of the popular ones are

1) Function Points 
2) Lines of Code 
3) Feature Points 
4) Mk II function points 
5) 3D Function Points 
6) Benchmarking 

We briefly explain each of the above methods in the next few pages

Function Points 
The Function Point methodology was developed by Allan Albrecht at IBM. This methodology is based on the belief that the size of a software project can be estimated during the requirements analysis. It takes into account the inputs and outputs of the system. Five classes of items are counted:

1. External Inputs 
2. External Outputs 
3. Logical Internal Files 
4. External Interface Files 
5. External Inquiries

The Total Function Point count is calculated based on the

a) Counts for each of these items 
b) The weighting factors and adjustment factors in this methodology

What are function points and why count them? 
“Function points are a measure of the size of Software applications and the projects that build them. The size is measured from a functional, or user, point of view. It is independent of the computer language, development methodology, technology or capability of the project team used to develop the application.”

Function points are not a perfect measure of effort to develop an application or of its business value, although the size in function points is typically an important factor in measuring each. Since the function point count for an application is independent of the technology used to develop the application it can be used for almost all types of applications such as GUI, OOP, Client Server, etc.
Since function points are based on screens, reports and other external objects, this measure takes the users' view. In these days of outsourcing and other confusion regarding the role of IT in an organization, understanding the users' view is of critical importance!

Lines of code: 
Counting lines of code measures software from the developers' point of view The number of lines of code is the traditional way of measuring the application size. Many people consider this method as irrelevant now. There are technical problems with the lines of code measure. It is difficult to compare lines of code when a mix of technologies is used. There is no standard definition of what a line of code is. A Program may have blank lines, comments, data declarations, and multi-line statements.

Feature points Methodology: 
It was developed by Software Productivity Research (SPR) in 1986. This technique takes into account the number of algorithms used in the application. It is compatible with the Function Points Methodology. The size calculated by the two methods for an ordinary transactional program would be the same. Feature Points Methodology is generally more useful for estimation in real-time process control, mathematical optimization and various embedded systems. The estimates are higher and considered more accurate in these cases.

Mk II function points Methodology: 
This was developed Charles Symons in 1984 at Nolan, Norton & Co., part of KPMG Management Consulting. The Original Function Point approach suffers from the following weaknesses:

· It is often difficult to identify the components of an application. 
· The original Function Point Methodology assigned weights to function point 
..components based on "debate and trial." 
· The original Function Point Methodology did not provide a means of accounting
..for internal complexity. ‘Feature points’ technique addresses these issues. 
· When small systems are combined into larger applications. Function 
..Points Methodology makes the total function point count less than the sum 
..of the components.

MKII decomposes the application being counted into a collection of logical transactions. Each transaction consists of an input, a process and an output. For each transaction, Unadjusted Function Points (UFP) become a function of the number of input data element-types, entity-types referenced and output data element-types. The UFPs for the entire system are then summed. Mk II is widely used in the UK, India, Singapore, Hong Kong and Europe. Users include governmental organizations, finance, insurance, retail and manufacturing.

3D function points: This methodology was developed by Boeing Company and published in 1992. The new technique was designed to address two classic problems associated with the Albrecht approach( the original Functional Point Methodology)

a) The original Functional Point Methodology is not user friendly 
b) It is inaccurate when measuring complex scientific and real-time systems.

The 3D function points takes into account the following Dimensions - data, function and control. The data dimension is similar to the original Function Point Methodology. The function dimension accounts for the transformations or algorithms. The control dimension accounts for transitions or changes in application state.

Benchmarking: Over the years many Organizations with significant development experience and mature processes have collected metrics on the various software development projects. These include the time, effort required to develop applications on various platforms and in various Business Domains. Based on this data benchmarks are created.

Each new software module to be developed can be categorized using the

a) Number of inputs 
b) Number of outputs 
c) Number of transactions 
d) Algorithms 
e) Features of the module

Based on the above factors the module can be categorized for example as Simple, Medium or Complex. If it is too Complex you could express it in multiples of the above three categories. The baseline effort in terms of person-hours it takes for each category is predefined based on historical data/metrics for a similar platform. This figure can be improvised/refined over a period of time This can be correlated to an algorithm for calculating Car Insurance Premium. This is used to estimate the size and the effort needed for Software Development.