Scope of software metrics

Scope, Benefit, and Cycle of Software Metrics

Software metrics have tremendous scope in software development to understand and fulfill various aspects of the project process. Software metrics help in estimation of various aspects like:-

1. Cost – Estimate project cost includes its maintenance, research, and other typical expenditure associated with the project. Putnam's SLIM model, Boehm's COCOMO model, and Albrecht's function model are implied in various cost estimation models.  
2. Quality assurance – Different metrics are used to measure different aspects of software quality, especially code quality (line of code).
3. Size and Complexity – It demonstrates the code size and complexity at the macro level of projects.
4. Functionality – Software metrics follow a scheduled procedure of software projects that focus on functionality, a document produced, and estimated time utilization.
5. Verdict – Software metrics help predict defects in code (pre and post-release of code). Also, predict project success and project risk.

Benefits of software metrics in software engineering

Software metrics are used to measure the product's performance to provide efficiency in their software development procedure. Benefits of software quality metrics are as follows:

1. Proactive management strategy – Due to management strategy, all the problems related to the product are evaluated and prioritized. Objective risks are identified, accessed, and managed.
2. Reliability- It helps to measure the stability and degree of risk of failure of the software.
3. Reduce ambiguity – Software metrics provide objective information, which reduces the complexity of constrained software projects.
4. Decision-maker – It helps decision-makers make effective decisions to meet project objectives, product performance, and software optimization.
5. Software metrics help reduce and find out the cost of developed software and overtime in making the product.
6. Increase project performance – Selection of best metric technical and project managers will increase project performance.
7. Helpful in increasing return on investment and provide opportunities for improvement.
8. Performance- Increase the performance of the software by testing code and by using application performance monitoring.

Life - Cycle of testing metric

Software testing metric life cycle consist of four steps:

1. Assessment and Analysis

This is the first phase of the software testing metric lifecycle that defines the identification and selection of the right software quality metrics for the software product.

• Communication

In this, the manager should communicate to the management about the type of software quality metrics and its usage. Stakeholders and the testing team have the right to know about the data point captured for processing the metric.

• Evaluation and processing

This phase software development team calculates the metric value and verifies the captured data before exhibiting it to management and stakeholders.

• Result

It is the last stage in which all the software metrics captured data is estimated in a report. These reports are sent to the management so that developers, testing teams, and stakeholders can record the improvements.

Various software metrics are used for software development; therefore, it becomes necessary to choose the correct metric for the software. So there are steps that define how to choose the correct metric for software development.

Step 1: Identify customers for each metric.

It is the first step where customers are identified for each metric. The customer is the person who require the metric data. Different types of customers have different information requirement; therefore, the customer plays a different role in different aspect of software development like:

• Testers
• Specialists
• Customers
• Software programmers
• Function and project management

Without the customer, the metric cannot be produced. The involvement of the customer in metric definition increases the chance of success.

Step 2: Select one or more goals.

In this second step, goals are selected for setting up the metric programs. Goals are selected after considering the different levels. Like at the organizational level, high-level goals are selected. High-level goals include a high level of customer satisfaction, profit margin target, or meeting project revenue. At the project level, goals like project level requirement and objective, control issues, project management are set by management. Based on organizational, task, and project goals, all the metrics are selected to meet these goals.

Step 3: Define questions.

It is the third step of defining the question that is answered to obtain the Goal. For example, Goal is to manufacture defect-free software, then the type of questions are asked:-

• Is the product is tested?
• Do all defects are removed?
• How many number of undetected defects are present?

Step 4: Select the required metric.

In this 4^th stage, such metrics are elected that can answer all the previous stage questions. During the selection of metric developer must be practical, pragmatic, and realistic. Metric does not solve the problem. It acts as an indicator that provides information based on which people can make an informed and intelligent decision. Metric help in understanding, evaluating, controlling, and predicting the future of software products, process, and services.

Step 5: Definitions for attributes and entries

This fifth step provides the standard definition for entries and their attributes. People interpret various ambiguous terms like quality, default, maintainability, report, problem size, and even project in their context, which creates misunderstanding. Therefore a standard definition is initiated. Terms like problem report, incident report, customer call report are meant the same for the people. To complain about a defect in software or any other default, the customer uses the problem to report or customer call report option. But these are different terms, so it becomes necessary to provide a standardized definition to terminologies to avoid barriers to understanding.

Step 6: Select measurement function for metrics

In this step, measurementfunctions are chosen to calculate the metric. Some base measure metrics directly measure the function consists of a single variable. More complex metrics, know as derived metrics, are modeled, for example, inspection's preparation rate, modeled as the number of lines of code. It is essential to select a model to use a measurement function. There are two methods.

• Use an existing model.
• Create a new one

Step 7: Mapping system- measurement method

In this seventh step of designing a metric, the function breaks down into its lowest level base measures, and values are assigned to those base measures. The mapping system is defined by the measurement method, and with this mapping system, numbers and symbols are assigned to the attributes. Some standardized units of measure, i.e., inches, liters, dolors, feet, pounds, help establish measurement methods. Base measure and its method provide the first level of data need to implement the metric. Data is collected by using models, and collected data is installed on each site.

Step 8: How metric is reported?

This eighth step defines the formatting of the data report. How all the collected data is reported in which format, extracted, define reporting mechanism, availability, and distribution. The report includes how the report looks like, i.e., data is arranged in tables, pie chart, line, graph, and bar charts.  Matric is delivered in the form of a hardcopy report, online electronic data, and email). The data extraction cycle defines how collecting data and the reporting cycle define the distribution of data in a report. The report is the medium to access the metric.   

Step 9: Determine metric qualifiers.

The ninth step is determining the metric qualifier when designing a metric.  Demographic information is provided by additional qualifiers needed at the second level of data collection requirement. Additional qualifier is a part of the metric design and foremost reason to define primitive data required.

Step 10: Collection of data

As we know, data is the root for metric, so it's essential "who should collect data"? Data "owner" is responsible for generating the data. If a person is collecting data, then it increases data accuracy (without anomalies) and completeness. Management provides time and resources to the people for data collection activities. Data analysis is based on collected data; therefore, data collection should be objective, assessable, convenient, and unambiguous. Data is recorded in a spreadsheet and other formats. Metric is very expensive; therefore, data collection must be accurate and recorded by its owner to reduce expenses.    

Step 11: Record people’s data

Selection of metric is dependent on the attitude of people involved in collecting data as measurement affect people and vice-versa. Measurement affects the individual's behavior as who is being measured; it is assumed to have importance. Metric is the result of human behavior, so this must be avoided by following some basic rules:

• Do not focus on people; just focus on processes and products.
• The first time metric is used against an individual; it provides valid data.
• Ignorance of data will surely kill a metric program.
• Do not use a single metric for measurement because software is complex, and with a single metric, all the customer's needs cannot be adequately measured.