?
0 Introduction I think I have to improve the quality of the development of software developers, there must be a corresponding assessment system. After the system, it can promote the development of developers to improve their own development quality. Although many companies have this performance appraisal, most of them do not have fine-grained metrics for software development, so they cannot assess the work performance of developers in accordance with effective metrics, most of them are only subjective will To assess, it cannot form an incurable persuversion to the assessor. The following documents are metrics and assessment methods written in accordance with previous experience and related information, hoping to use the company's improvement assessment system. Due to the limited time, there is deficiencies, please contact your advice, thank you! 1 Objective To quantify the quality and process of software items generated by software development, and use the results of the evaluation to guide the development process of software, constantly improve software development quality, and assess the work of software developers according to metrics. Performance. 2 Software items include 1) Technical documentation: Refer to the company "Software Engineering Product Set", the determined configuration item; mainly include: feasibility analysis report, demand analysis report, software function specification, development plan, quality plan, system design report , Test documentation, technical report, user manual, summary report, etc .; 2) computer program. 3 Source 1) Project Plan; 2) Review Report; 3) Test Report; 4) Problem Report; 5) Software Maintenance Records; 4 Quality Measurement 4.1 Metrics Mainly based on the inspection indicators of various software item checklists For example, software requirements Specifications checklist (see Appendix 1), there are 10 inspection indicators, depending on the specific project check side, the corresponding check indicator can be selected as a metrics. 4.2 Quality Level 1) The determination of the quality level of the software item is performed according to the metric synthesis index. 2) The calculation formula for metrics is: Total = σqimi. 3) Where i = 1, 2, ... n represents the number of indicators; 4) q represents metric indicators; 5) M represents the measured index q, the weight coefficients accounted for the entire indicator system, may be different development projects Different, this coefficient is given in accordance with the development of the development. Metric index weights table: serial number indicator weight 1 indicator 1 Right 12 indicator 2 Rights 23 Indicator 3 Right 34 Indicator 4 Right 45 Indicator 5 Right 5 Weighted Average Siza 1.06) Quality Evaluation: General, according to Metrics Comprehensive Indicators Value, there is the following score standard. Quality evaluation score standard number sequence number score quality evaluation 1 9.0 ~ 10.0 high quality 2 8.0 ~ 8.9 Good 3 6.0 ~ 7.9 Qualified 4 below 6.0 Inferior 4.3 Metrics Calculation Method 1) Metrics Score Standard: According to the inspection indicators of the software item The defect rate is determined that the score standard is made for each metric index, the higher the defect rate, the lower the score. Metric Score Standard Size Sequence Score Defect Rate Rate 1 9 ~ 10 Range 13 7 7 ~ 8 Range 34 5 ~ 6 Range 45 below 5 Range 52) Defect Rate Source: Mainly Software Item Check, Review, Test Process Office The resulting defect trace table (see Appendix 2), the defect tracking table in the defect class corresponds to the inspection indicator in the table. 3) Defect Efficiency Units <1> Structured Method Demand Document: Number of function points or documentation (number of defects / function points, number of defects / pages). Design documentation: units in modules or documentation (defects / modules, defects / pages).
Software code: Unit as modules or code lines (number of defects / modules, number of defects / LOC). <2> Object-Oriented Method Demand Document: Unit with business use cases or document pages (number of defects / BUC, defects / pages). Design documentation: in units of use cases or documentation (number of defects / UC, defects / pages). Software code: use case or code line number (number of defects / UC, defects / LOC). 4) Defect rate calculation method: <1> Firstly, according to the defects in the defect tracking table, various types of defect rates are respectively statistically, and then points to each metric score based on the metric score standard table. <2> Defect rate calculation formula is: Total = σ (Bi / U) MI. <3> where i = 1, 2, ... n represents the defect level, the defect level can be divided into: fatal, severe, general, slight; <4> b represents the number of defects of each defect; <5> u The number of defects; <6> M represents the weight coefficient of each defect level, usually the weight coefficient is based on the "general" defect level as the base (the right number is 1), "slight" defective level can not calculate the defect rate (power) Set to 0). Defect level weight coefficient table: Sequence level power 1 fatal 32 severe 23 General 14 Mimi 05 Overprocess metric 5.1 Time Dance 1) The completion time based on software project plans is based, and the time difference between the actual completion time and the scheduled completion time. The ratio of the software item is completed to evaluate the timely time of each software item. 2) Time difference = (actual completion time - plan completion time) / (Plan start time - plan completion time) 3) The scoring criteria are as follows: The time difference rate range can be determined according to specific items. Timely score standard number sequence number score time difference range 1 9 ~ 10 Range 1 (within 15%) 3 7 ~ 8 Range 3 (15% ~ 35%) 4 below 7 Range 4 (35% or more) 5.2 maturity The results of the item inspection, the review, and the test are the evaluation benchmark. The scoring criteria are as follows: <1> 9 ~ 10 points: one inspection, review, test, no need to adjust; <2> 7 ~ 8 points: one check, review, test Slightly adjust, or the second inspection, review, and test pass without adjustment; <3> 6 points below: Temporary inspection, review, test is not passed. 5.3 Improve efficiency In the examination, review, and testing results, improve software item results, in improved time affects the completion of the completion of subsequent stages and the overall completion time of the plan, the score standard is as follows: <1> 9 ~ 10 points : Improved efficiency, the modification of software items does not have to increase workload, and does not affect the continued period of time and the overall completion of the project plan, or no need to be revised; <2> 7 ~ 8 points: improve efficiency, modification or improvement of software items The continuation of the stage, the increase of workload is 20% in the original plan (this threshold can be determined according to specific items); <3> 6 points below: The improvement efficiency is poor, the modification or improvement of the software item makes the project extension, or The increased workload exceeds 20% (this threshold can be determined according to specific items). 5.4 Comprehensive Evaluation 1) The comprehensive score method for process metrics is: s = p * a1 m * A2 E * A3.
2) S represents a comprehensive score; 3) P representation time; 4) M represents maturity; 5) E represents improved efficiency; 6) A representative process indicators (P, M, E) occupy the weight in the entire indicator system The coefficient (depending on the specific item). Process indicator weight restraining table: Sequence number indicator weight 1 time time number 1 (0.3) 2 Maturity weight 2 (0.3) 3 Improve efficiency power 3 (0.4) Weighted average score 1.07) Process evaluation: General, according to process synthesis Rating, with the following score standard. Process evaluation score standard form: serial number score evaluation 1 9.0 ~ 10.0 high quality 2 8.0 ~ 8.9 good 3 6.0 ~ 7.9 qualified 4 below 6.0 inferior 6 Software Item Comprehensive Evaluation 1) Software Item Comprehensive Score = Quality Score * Quality Right Process score * process power. Comprehensive indicator Right Ratio Table: Sequence Number Right 1 Quality Right 1 (0.5) 2 Procedure 2 (0.5) Weighted Average 1.02) Software Item Comprehensive Evaluation: Generally, according to software item comprehensive score, there is the following score standard. Software Item Integrated Evaluation Score Standard Table: Sequence Score Process Evaluation 1 9.0 ~ 10.0 High Quality 2 8.0 ~ 8.9 Good 3 6.0 ~ 7.9 Qualified 4 below 6.0 Perit Media 7 Metric Record Project Manager or Development Manager Subscription Inspection, Review, Test After passing, you need to fill in the software item metric table (see Appendix 3) and submit to the SQA Department Audit and the Managers Manager Approved. If a software item is completed by multiplayer, you need to split the software item (function, module), record the metrics of each person, and the SQA department fills in the software item comprehensive evaluation table according to the software item metric. Appendix 4). 8 Performance Appraisal 8.1 Developer Assessment 1) The average score calculation method of software item is: AV = (COI) /i *10.2) AV represents the average score value of the software item; 3) CO represents software item synthesis Score; 4) i = 1, 2, ... n represents the number of software items. 5) Performance assessment standard: Generally, according to the average score of software items, there is the following score standard. Performance Assessment Score Standard Table Sequence Subscription Assessment Evaluation 1 90 ~ 100 Excellent 2 80 ~ 89 Good 3 60 ~ 79 and Grid 4 below 60 Very Wet 6) Software Departments According to Software Item Comprehensive Evaluation Forms Each month or quarterly developers The average score value of the software item, the average score of the developer software item and the scope of performance assessment standards are determined, and the development personnel performance evaluation is determined. The performance appraisal is "good", and the performance appraisal is "very poor" personnel punishment corresponding amount. For poor developers need to report criticism, and ask for the project manager and the development manager to write personal software development process improvement. If it is very poor for three consecutive months, it may be downgraded, wage, and even dismissal; for excellent developers to report, and organize experience exchanges, introduce its excellent software development process control methods, if it is three consecutive months It is excellent, it may be upgraded and saved. 8.2 Project Manager, Development Manager Issue 1) The average score calculation method of the project team is: AV = (σcoi) / i * 10.
2) AV represents the average score of the project group; 3) CO represents project group members performance assessment score; 4) i = 1, 2, ... n represents the number of members of the project group. 5) Performance assessment standard: Generally, according to the average score of the project group, there is the following score standard. Performance Assessment Score Standard Size Sequence Subspection Evaluation 1 90 ~ 100 Excellent 2 80 ~ 89 Good 3 60 ~ 79 and Grid 4 below 60 Very Wat 6) After the software department determines the project team member performance appraisal evaluation, Calculate the average score value of the project team, the average score of the project group and the scope of performance assessment standards, determine the project manager, and develop manager performance assessment. Performance assessment is "good" project manager, the development manager rewards the corresponding amount, and the performance assessment is "very poor" project manager, and the development manager's penalty will be punished. For poor project managers, the development manager needs to report criticism, and requires the improvement of the project group software development process with help of the department manager. If it is three consecutive months, it may be downgrade, reduce salary, and even dismissal; For excellent project managers, the development manager is notified, and the experience exchange will introduce its excellent project team development process management methods. If it is excellent for three consecutive months, it may be upgraded, and the salary will be upgraded. 8.3 Inspection, Test Personnel Appraisal 1) In software item inspection, review, testing, project managers or development managers need to fill in inspections, testers' defect lookup quality metrics (see Appendix 5) as a basis for monthly assessment or quarterly assessment basis. 2) Defect search rate calculation method <1> Defect lookup rate calculation formula is: total = σ (bi / u) mi. <2> where i = 1, 2, ... n represents the defect level, the defect level can be divided into: fatal, severe, general, slight; <3> B represents the number of defects of each defect; <4> u representative The number of defects; <5> M represents the weight coefficient of each defect level, usually the weight coefficient is based on the "general" defect level as the base (the right number is 1), "slight" defective level can not calculate the defect lookup rate (right) The number is set to 0). Defect level weight factor table: Sequence level power 1 fatal 32 severe 23 general 14 slight 03) Defect search quality evaluation: Develop appropriate defects to find quality score standards for each software item. Defect search quality score standard form: Sequence Score Defect Finding Rate Range 1 9 ~ 10 Range 13 7 ~ 8 Range 34 5 ~ 6 Range 45 below 5 Range 54) The average score calculation method of defect lookup is: AV = ΣCOI) / i * 10. <1> AV represents the average score of defect search; <2> CO represents defect lookup quality score; <3> i = 1, 2, ... n represents the number of software items. 5) Performance assessment standard performance assessment score standard table: serial number score assessment evaluation 1 90 ~ 100 excellent 2 76 ~ 89 Good 3 60 ~ 75 and 3 below 60 Very poor 6) Software departments According to software item comprehensive evaluation table each month Or quarterly statistics of the average score value of each examinator or tester defect lookup, the average score value of the test person or the test personnel lookup, the scope of the performance assessment standard, determine the performance appraisal evaluation of the inspection person or test personnel, performance assessment " Good "The above people reward the corresponding amount. For excellent staff, and organize experience exchanges will introduce its excellent review and test methods.
