CN105302724A - Instant defect predicting method based on mixed effect removing - Google Patents

Instant defect predicting method based on mixed effect removing Download PDF

Info

Publication number
CN105302724A
CN105302724A CN201510755374.9A CN201510755374A CN105302724A CN 105302724 A CN105302724 A CN 105302724A CN 201510755374 A CN201510755374 A CN 201510755374A CN 105302724 A CN105302724 A CN 105302724A
Authority
CN
China
Prior art keywords
change
tolerance
developer
sequence
amount
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201510755374.9A
Other languages
Chinese (zh)
Inventor
卢红敏
杨已彪
刘金平
赵泱泱
周毓明
徐宝文
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing University
Original Assignee
Nanjing University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nanjing University filed Critical Nanjing University
Priority to CN201510755374.9A priority Critical patent/CN105302724A/en
Publication of CN105302724A publication Critical patent/CN105302724A/en
Pending legal-status Critical Current

Links

Landscapes

  • Stored Programmes (AREA)

Abstract

The invention provides an instant defect predicting method based on mixed effect removing. The method comprises the following steps that 1, software change measurement is collected; 2, a mixed effect of the change measurement and change amount measurement is removed; 3, an unsupervised instant defect predicting model is built; 4, a change sequence is submitted to developers to be examined. According to the instant defect predicting method based on mixed effect removing, the problems that in an existing traditional instant defect predicting method, the data collecting difficulty is large, and the predicting capacity is insufficient are solved; not only can the instant defect predicting model be simply and effectively built, but also the examination cost can be significantly saved, therefore, limited resources are fully utilized, and the software product quality is better controlled and improved.

Description

A kind of instant failure prediction method removed based on mixed effect
Technical field
The invention belongs to field of software engineering, especially instant software defect prediction field, and relate to a kind of instant failure prediction method removed based on mixed effect especially.
Background technology
Along with the development of software engineering, user requires more and more higher to software quality, and software developer is also constantly seeking the method for raising software quality to meet consumers' demand.Software defect is problem, the mistake that certain that exist in computer software or program destroys normal service ability, or the functional defect hidden, and is the inefficacy of required certain function realized of system or runs counter to.Software defect, along with each process of software development flow, if demand analysis stage does not fully understand fully that demand will bring much unnecessary software defect, does not adopt outstanding management method can cause a lot of software defect yet in performance history.Examination and reparation software defect need to drop into a large amount of manpower and materials, and occupy the most of the time in software life-cycle.
Instant software defect forecasting techniques is a method being intended to the defect introduced in forecasting software performance history, help developer identifies that when code is submitted to the code may introducing defect is submitted to, distribute developer timely to examine the code submitted to targetedly, effectively distribute limited resource, identify the code introducing defect fast submit to and repair, thus better control and promote the quality of software product.
Summary of the invention
The object of the invention is to provide a kind of instant failure prediction method removed based on mixed effect, solve the instant failure prediction method defective data collection difficulty existed at present large, the problem of estimated performance difference, instant software defect forecasting techniques is made to have adaptability widely, promote accuracy rate and the precision of this technology simultaneously, the workload of reasonable distribution examination, improves the efficiency of code inspection, thus promotes the quality of software product.
For reaching above-mentioned purpose, the present invention proposes a kind of instant failure prediction method removed based on mixed effect.Method comprises the following steps:
1) collection of tolerance is changed: version control system have recorded the information (comprise this submission and have modified which file, the information such as submitter, time and daily record) that in software development process, all history is submitted to.According to these historical informations recorded in version repository, excavate and change tolerance.Here, a change refers to submission set all in given interval.Change tolerance and comprise increase, delete code amount, amount of change (namely increase and delete ground total code line number), the file number be modified, subsystem number, the information such as submitter's number that change relates to and experience;
2) change tolerance and mix removing of effect with amount of change: first, set up the linear regression model (LRM) changed between tolerance and amount of change, then measure changing the predicted value that original value deducts linear regression model (LRM), remove with this mixed effect changed between tolerance and amount of change, thus obtain new change tolerance.
3) change by the metric sequence after removing mixed effect: utilize in step 2 the new change metric removing mixed effect and obtain, from small to large change is sorted, obtain defect and introduce and change tendentious sequence;
4) developer examines successively according to the order of sequence and repairs change: change sequence step 3 obtained submits to developer, and developer examines change according to the order of sequence successively, differentiates that whether change is that defect introduces change, if so, then repairs this change.
Further, wherein the concrete steps of above-mentioned steps 1 are as follows:
Step 1-1: initial state;
Step 1-2: from version repository, obtains the identifier submitted to;
Step 1-3: according to the identifier submitted to, obtains the relevant information of this submission of recording in version repository;
Step 1-4: by given interval, divide into groups to submitting in version repository, one changes to one group of submission;
Step 1-5: according to submission grouping information, software for calculation changes tolerance;
Step 1-6: change tolerance and collect complete.
Further, wherein the concrete steps of above-mentioned steps 2 are as follows:
Step 2-1: initial state;
Step 2-2: to each change tolerance c, differentiates whether be amount of change tolerance;
Step 2-3: if c is not amount of change tolerance, set up the linear regression model (LRM) C=β S+ ε between c and amount of change s;
Step 2-4: predicted value p (the c)=β s+ ε deducting regression model from c;
Step 2-5: obtain new change metric c '=c-p (c);
Step 2-6: it is complete that change tolerance and amount of change mix removing of effect.
Further, wherein the concrete steps of above-mentioned steps 3 are as follows:
Step 3-1: initial state;
Step 3-2: the new change tolerance obtained based on step 2, chooses suitable change tolerance c ';
Step 3-3: to sort from small to large change by c ';
Step 3-4: obtain the change sequence sorting and obtain;
Step 3-5: change complete by the metric sequence removed after mixed effect.
Further, wherein the concrete steps of above-mentioned steps 4 are as follows:
Step 4-1: initial state;
Step 4-2: the change sequence that developer obtains by step 3 examines change successively;
Step 4-3: if change to defect to introduce change, repair and change more;
Step 4-4: it is complete that developer examines and repair change according to the order of sequence successively.
The present invention sets up instant bug prediction model by removing the mixed effect changed between tolerance and change size of code, defective data collection is not needed to carry out training pattern, therefore compared to the current instant bug prediction model having supervision, not only do not need to collect defect information, significantly can also improve the performance of instant bug prediction model simultaneously.Therefore, this method has the practicality of adaptability and Geng Gao widely.Based on the change defect tendentiousness sequence that obtains of forecast model that this method is set up, developer can examine change more targetedly, thus can effectively utilize and exploit natural resources, and significantly improves the quality of software product.
Accompanying drawing explanation
Fig. 1 is the process flow diagram of a kind of instant failure prediction method based on mixed effect model of the embodiment of the present invention.
Fig. 2 is the process flow diagram changing tolerance collection in Fig. 1.
Fig. 3 changes the process flow diagram removed measured and mix effect with amount of change in Fig. 1.
Fig. 4 measures the process flow diagram sorting and change in Fig. 1 after removing mixed effect.
Fig. 5 is the process flow diagram that in Fig. 1, developer's examination and reparation are changed.
Embodiment
For understanding technology contents of the present invention, institute's accompanying drawings is coordinated to be described as follows especially exemplified by specific embodiment.
Fig. 1 is the process flow diagram of a kind of instant failure prediction method removed based on mixed effect of the embodiment of the present invention.Based on the instant failure prediction method that mixed effect removes, it is characterized in that comprising the following steps:
S101 changes the collection of tolerance: version control system have recorded the information (comprise this submission and have modified which file, the information such as submitter, time and daily record) that in software development process, all history is submitted to.According to these historical informations recorded in version repository, excavate and change tolerance.Here, a change refers to submission set all in given interval.Change tolerance and comprise increase, delete code amount, amount of change (namely increase and delete ground total code line number), the file number be modified, subsystem number, the information such as submitter's number that change relates to and experience;
S102 changes tolerance and mixes removing of effect with amount of change: first, set up the linear regression model (LRM) changed between tolerance and amount of change, then measure changing the predicted value that original value deducts linear regression model (LRM), remove with this mixed effect changed between tolerance and amount of change, thus obtain new change tolerance;
S103 changes by the metric sequence after removing mixed effect: utilize in step 2 the new change metric removing mixed effect and obtain, sort from small to large to change, obtains defect and introduces and change tendentious sequence;
S104 developer examines successively according to the order of sequence and repairs change: change sequence step 3 obtained submits to developer, and developer examines change according to the order of sequence successively, differentiates that whether change is that defect introduces change, if so, then repairs this change.
Fig. 2 is the process flow diagram changing tolerance collection.Version control system have recorded the information (comprise this submission and have modified which file, the information such as submitter, time and daily record) that in software development process, all history is submitted to.According to these historical informations recorded in version repository, excavate and change tolerance.Here, a change refers to submission set all in given interval.Change tolerance and comprise increase, delete code amount, amount of change (namely increase and delete ground total code line number), the file number be modified, subsystem number, the information such as submitter's number that change relates to and experience.Concrete steps are as follows: step 1: initial state; Step 2: from version repository, obtains the identifier submitted to; Step 3: according to the identifier submitted to, obtains the relevant information of this submission of recording in version repository, as the daily record of submission time, submitter, submission and the content etc. of amendment; Step 4: by given interval, divide into groups to submitting in version repository, one changes to one group of submission; Step 5: according to submission grouping information, software for calculation changes tolerance; Step 6: change tolerance and collect complete.
Fig. 3 removes the process flow diagram changing and measure and mix effect with amount of change.Remove by the method for linear regression the mixed effect changing tolerance and code variation and obtain new tolerance.Concrete steps are as follows: step 1: initial state; Step 2: to each change tolerance c, differentiates whether be amount of change tolerance; Step 3: if c is not amount of change tolerance, set up the linear regression model (LRM) C=β S+ ε between c and amount of change s; Step 4: predicted value p (the c)=β s+ ε deducting regression model from c; Step 5: obtain new change metric c '=c-p (c); Step 6: it is complete that change tolerance and amount of change mix removing of effect.
Fig. 4 is the process flow diagram by measuring after removing mixed effect changing sequence permutation.According to the change tolerance removed after mixed effect, by sorting to change introducing from small to large.Concrete steps are as follows: step 1: initial state; Step 2: the new change tolerance obtained based on step 2, chooses suitable change tolerance c '; Step 3: to sort from small to large change by c; Step 4: obtain the change sequence sorting and obtain; Step 5: change complete by the metric sequence removed after mixed effect.
Fig. 5 is that developer examines the process flow diagram changing sequence.Change sequence step 3 obtained submits to developer, and leu time examination is changed and identified whether as defect changes according to the order of sequence, and if so, repair-deficiency changes.Concrete steps are as follows: step 1: initial state; Step 2: the change sequence that developer obtains by step 3 examines change successively; Step 3: if change to defect to introduce change, repair and change more; Step 4: it is complete that developer examines and repair change according to the order of sequence successively.
In sum, the invention solves the problem that collection data difficulty is large, predictive ability is not enough that the instant failure prediction method of tradition at present exists, not only can the instant bug prediction model of simple and effective foundation, can also significantly save examination cost simultaneously, thus make full use of limited resource, better control and improve speed measuring with software.

Claims (5)

1. the instant failure prediction method removed based on mixed effect, it is characterized in that, by setting up the linear regression model (LRM) changed between tolerance and amount of change, use the original value changing tolerance to deduct the method changing measurement model predicted value and remove the mixed effect changed between tolerance and amount of change, new change tolerance is finally used to sort to change, then change sequence is submitted to developer to examine, change to identify defect introducing and repair, can effectively distribute limited resource like this, developer is made to examine code targetedly, Timeliness coverage repair-deficiency, thus better control and promote the quality of software product, the method comprises the following steps:
1) collection of tolerance is changed;
Definition 1: submission is the historical information that in version control system, developer submits to code revision to record, comprises information such as submitting identifier, submission date, developer and influenced file to;
Definition 2: change is one group of set of all submissions in given interval;
Version control system have recorded the information that in software development process, all history is submitted to, according to the historical information recorded in version repository, excavate and change tolerance, change tolerance and comprise increase, delete code amount, amount of change (namely increase and delete ground total code line number), the file number be modified, subsystem number, the information such as the developer's number involved by change and developer's experience;
2) change tolerance and mix removing of effect with amount of change; First, set up the linear regression model (LRM) changed between tolerance and amount of change, then measuring changing the predicted value that original value deducts linear regression model (LRM), removing with this mixed effect changed between tolerance and amount of change, thus obtain new change tolerance;
3) change by the metric sequence after removing mixed effect: utilize in step 2 the new change metric removing mixed effect and obtain, from small to large change is sorted, obtain defect and introduce and change tendentious sequence;
4) developer examines successively according to the order of sequence and repairs change: change sequence step 3 obtained submits to developer, and developer examines change according to the order of sequence successively, differentiates that whether change is that defect introduces change, if so, then repairs this change.
2. the instant failure prediction method based on mixed effect model according to claim 1, is characterized in that, in step 1) in, change the collection of tolerance; Version control system have recorded the information that in software development process, all history is submitted to, according to the historical information recorded in version repository, excavate and change tolerance, change tolerance and comprise increase, delete code amount, amount of change (namely increase and delete ground total code line number), the file number be modified, subsystem number, the information such as the developer's number involved by change and developer's experience.
3. the software defect automatic classification based on abstract syntax tree according to claim 1 and software quality real-time control method, is characterized in that, in step 2) in, change tolerance and mix removing of effect with amount of change; First, set up the linear regression model (LRM) changed between tolerance and amount of change, then measuring changing the predicted value that original value deducts linear regression model (LRM), removing with this mixed effect changed between tolerance and amount of change, thus obtain new change tolerance.
4. the software defect automatic classification based on abstract syntax tree according to claim 1 and software quality real-time control method, it is characterized in that, in step 3) in, change by the metric sequence after removing mixed effect: utilize in step 2 the new change metric removing mixed effect and obtain, from small to large change is sorted, obtain defect and introduce the tendentious sequence of change.
5. the software defect automatic classification based on abstract syntax tree according to claim 1 and software quality real-time control method, it is characterized in that, in step 4) in, developer examines successively according to the order of sequence and repairs change: change sequence step 3 obtained submits to developer, developer examines change according to the order of sequence successively, differentiate that whether change is that defect introduces change, if so, then repairs this change.
CN201510755374.9A 2015-11-05 2015-11-05 Instant defect predicting method based on mixed effect removing Pending CN105302724A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201510755374.9A CN105302724A (en) 2015-11-05 2015-11-05 Instant defect predicting method based on mixed effect removing

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201510755374.9A CN105302724A (en) 2015-11-05 2015-11-05 Instant defect predicting method based on mixed effect removing

Publications (1)

Publication Number Publication Date
CN105302724A true CN105302724A (en) 2016-02-03

Family

ID=55200013

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201510755374.9A Pending CN105302724A (en) 2015-11-05 2015-11-05 Instant defect predicting method based on mixed effect removing

Country Status (1)

Country Link
CN (1) CN105302724A (en)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106502909A (en) * 2016-11-07 2017-03-15 南京大学 A kind of aacode defect Forecasting Methodology in smart mobile phone application and development
CN108170466A (en) * 2017-12-21 2018-06-15 南京大学 A kind of C/C++ bugs self-repairing methods based on program synthesis
CN106021115B (en) * 2016-06-06 2018-07-10 重庆大学 Unsupervised failure prediction method based on probability
CN108614778A (en) * 2018-05-10 2018-10-02 天津大学 Prediction technique is changed based on the Android App program evolutions that Gaussian process returns
CN108763063A (en) * 2018-05-09 2018-11-06 南京大学 A kind of software defect detection method without defect labeled data
CN111611010A (en) * 2020-04-24 2020-09-01 武汉大学 Interpretable method for code modification real-time defect prediction

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103678091A (en) * 2013-12-19 2014-03-26 北京奇虎科技有限公司 Method and device for processing crash data of application software

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103678091A (en) * 2013-12-19 2014-03-26 北京奇虎科技有限公司 Method and device for processing crash data of application software

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
ZHOU YUMING 等: "An in-depth study of the potentially confounding effect of class size in fault prediction", 《ACM TRANS. ON SOFTWARE ENGINEERING AND METHODOLOGY》 *

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106021115B (en) * 2016-06-06 2018-07-10 重庆大学 Unsupervised failure prediction method based on probability
CN106502909A (en) * 2016-11-07 2017-03-15 南京大学 A kind of aacode defect Forecasting Methodology in smart mobile phone application and development
CN106502909B (en) * 2016-11-07 2019-04-23 南京大学 A kind of aacode defect prediction technique in smart mobile phone application exploitation
CN108170466A (en) * 2017-12-21 2018-06-15 南京大学 A kind of C/C++ bugs self-repairing methods based on program synthesis
CN108170466B (en) * 2017-12-21 2019-09-20 南京大学 A kind of C/C++ bugs self-repairing method based on program synthesis
CN108763063A (en) * 2018-05-09 2018-11-06 南京大学 A kind of software defect detection method without defect labeled data
CN108763063B (en) * 2018-05-09 2022-07-12 南京大学 Software defect detection method without defect labeling data
CN108614778A (en) * 2018-05-10 2018-10-02 天津大学 Prediction technique is changed based on the Android App program evolutions that Gaussian process returns
CN111611010A (en) * 2020-04-24 2020-09-01 武汉大学 Interpretable method for code modification real-time defect prediction
CN111611010B (en) * 2020-04-24 2021-10-08 武汉大学 Interpretable method for code modification real-time defect prediction

Similar Documents

Publication Publication Date Title
CN105302724A (en) Instant defect predicting method based on mixed effect removing
CN111103565B (en) Data transformation method and system based on intelligent electric energy meter metering error analysis
CN104572449A (en) Automatic test method based on case library
CN102402726A (en) Method for predicting electric quantity of large-scale distribution network based on regional load analysis
CN109189479A (en) A kind of parallel automatic verification method for processor instruction set
CN104680254B (en) A kind of power network planning scheme method for optimizing based on integrated cost model
CN104881715B (en) A kind of method that paper plant predicts pulp property according to waste paper proportioning
CN104252652A (en) Space load predicting method in electricity system
CN115079081B (en) Intelligent ammeter metering abnormality identification method and device based on electricity consumption data
CN109324978A (en) A kind of software testing management system of multi-person synergy
CN107766500A (en) The auditing method of fixed assets card
CN111177128B (en) Metering big data batch processing method and system based on improved outlier detection algorithm
CN114511181A (en) Water pollution environment-friendly calibration method and device based on power grid and tax data fusion
CN110968703B (en) Method and system for constructing abnormal metering point knowledge base based on LSTM end-to-end extraction algorithm
CN110287114B (en) Method and device for testing performance of database script
Wang et al. A review of a series of effective methods in urban metabolism: Material flow, ecological network and factor analysis
CN110490220A (en) A kind of bus load discrimination method and system
CN106227647A (en) A kind of data analysing method for server hardware state evaluation
CN113793057A (en) Building bidding and tendering data generation method based on regression analysis model
Yu et al. Research on the post occupancy evaluation of green public building environmental performance combined with carbon emissions accounting
CN108805742B (en) Method and device for evaluating electric energy substitution potential of user
CN107194529B (en) Power distribution network reliability economic benefit analysis method and device based on mining technology
CN109871998B (en) Power distribution network line loss rate prediction method and device based on expert sample library
CN103530805A (en) Automatic settling system and method based on database and system interface
Guo et al. A multi-event combination maintenance model based on event correlation

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
WD01 Invention patent application deemed withdrawn after publication
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20160203