Function-oriented software measurement

Function-oriented Measurement

From: http://www.zxbc.cn/html/rjgc/2019251939768.html

Function-oriented software measurements are indirect measurements of software and software development processes. Function-oriented measurements focus on the program's "functionality" and "Practicality", rather than counting loc. This metric was first proposed by Albrecht. He proposed a productivity measurement method called the function point method, which utilizes empirical relationships between counting measurement and software complexity Estimation in software data domains, export Function Point FPS (function points ). The function is calculated by entering the chart format. First, identify the features of the five data domains and give a count in the corresponding position in the table. The data field value is defined as follows: (1) number of user inputs: each user input should be input data for different applications and be counted. The input data should be different from the query data, and the numbers of juice should be different. (2) user output count: each user output is the application-oriented output information provided to the user, which should be counted. The "output" here refers to reports, screen information, error information, and so on. Each data item in the report should not be counted separately. (3) user query count: a query is an online input that triggers the software to generate an instant response in the online mode. Each query must be counted. (4) number of files: each logical master file should be counted. The logical master file here refers to a set of data in the logic. It can be a part of a large database or a separate file. (5) Number of External Interfaces: All interfaces used to transmit information to another system (that is, data files on tapes, disks, and read/write discs) should be counted. Once the above data is collected, the weighted complexity values related to each count can be calculated. The unit that uses the function points method should develop some guidelines to determine whether a specific item is simple, average, or complex. However, the determination of complexity still involves subjective factors. Calculate the function using the following relationship: Fp = total × (0.65 + O.01 × sum (FI) (13.1) Where, the sum is the sum of all weighted complexity values; FI (I = 1 to 14) is the complexity correction value, which should be determined by answering the questions one by one. Sum (FI) is the sum function. The constants in the preceding equation and the weighting factor applied to the data domain count can be determined by experience. Once the function points are calculated, the productivity, quality, and other attributes of the software can be measured in the way of LOC: Productivity = FP/PM (person-month) Quality = number of errors/FP Cost = RMB/FP Document = document pages/FP The extended feature point measurement is called the feature point FPS (feature points) measurement. It is suitable for applications with high algorithm complexity. However, real-time processing, process control, and embedded software applications have high algorithm complexity. Therefore, it is suitable for special point measurement. To calculate the feature points, you can count and weight the data domain values as described above. In addition, feature point measurement also counts a new software feature "Algorithm. It can be defined as a bounded computing problem contained in a specific computer program ". For example, matrix inversion, binary string conversion to decimal number, and processing an interrupt are all algorithms. You can use tables to calculate feature points. Only one weight is used for each measurement parameter, and the equation (13.1) is used to calculate the total feature point value. It must be noted that feature points and feature points represent the same thing: "functionality, or" Practicality "obtained by the software ". In fact, for traditional engineering computing or information system applications, the two measurements will produce the same FP value. In a complex real-time system, the Count of feature points is usually 20% to 35% higher than that determined by only functional points. Like Loc, the measurement of function points (or feature points) is controversial. Advocates believe that FP is irrelevant to programming languages and is ideal for applications using traditional and non-procedural languages; it is also based on data that may be known early in project evaluation. Therefore, FP is attractive as an estimation method. Opponents believe that this method relies on subjective factors rather than objective reality in its calculation. Data in the data field is difficult to collect afterwards, and FP has no direct physical meaning. It is just a number.