Predicting Defects using Network Analysis on Dependency Graphs
4 likes2,243 views
Network analysis of dependency graphs between code components can help predict defects and prioritize testing. Analyzing dependencies captured metrics like centrality and ego networks that strongly correlated with past defects. These network measures improved prediction of defects, ranking highest risk binaries, and identification of critical "escrow" binaries requiring extra testing over complexity metrics alone, doubling recall of escrow binaries. Combining network and complexity analyses provides more effective defect prediction and testing prioritization to aid new QA managers.
1 of 61
Ad
Recommended
Bicliques for Preimages: Attacks on Skein-512 and the SHA-2 family
Bicliques for Preimages: Attacks on Skein-512 and the SHA-2 familyAlexandra Bicliques for Preimages: Attacks on Skein-512 and the SHA-2 family
by Dmitry Khovratovich, Christian Rechberger,
Alexandra Savelieva.
Presented at 19th International Workshop on Fast Software Encryption - FSE 2012 (March 2012, Washington DC)
Changes and Bugs: Mining and Predicting Development Activities
Changes and Bugs: Mining and Predicting Development ActivitiesThomas Zimmermann The document discusses mining software archives and bug databases to predict software development activities and defects. It presents two main contributions: 1) fine-grained analysis of version archives to identify usage patterns and cross-cutting changes, and 2) mining bug databases to predict defects based on dependencies and the increased likelihood of defects when depending on defect-prone code.
Multiple Choice Questions on JAVA (object oriented programming) bank 7 -- abs...
Multiple Choice Questions on JAVA (object oriented programming) bank 7 -- abs...Kuntal Bhowmick The document contains a collection of multiple choice questions and answers about abstract classes in object-oriented programming. It includes 14 questions that test understanding of key concepts of abstract classes such as: when the abstract keyword is used, defining abstract methods, preventing instantiation of abstract classes, requiring subclasses to implement abstract methods, and preventing inheritance of classes. Each question is presented on an even page with its answer on the adjacent odd page.
Multiple Choice Questions on JAVA (object oriented programming) bank 6 -- inh...
Multiple Choice Questions on JAVA (object oriented programming) bank 6 -- inh...Kuntal Bhowmick This document contains a 20 question multiple choice quiz about object oriented programming concepts related to inheritance. Each question is presented on an even page with the corresponding answer and explanation on the adjacent odd page. The quiz covers topics like superclass vs subclass, method overriding vs hiding, inheritance terminology in Java, and advantages/disadvantages of inheritance. The document instructs readers to first attempt each question before looking at the solution, and suggests viewing it in single page mode for clarity.
Belief Propagation Decoder for LDPC Codes Based on VLSI Implementation
Belief Propagation Decoder for LDPC Codes Based on VLSI Implementationinventionjournals International Journal of Engineering and Science Invention (IJESI) is an international journal intended for professionals and researchers in all fields of computer science and electronics. IJESI publishes research articles and reviews within the whole field Engineering Science and Technology, new teaching methods, assessment, validation and the impact of new technologies and it will continue to provide information on the latest trends and developments in this ever-expanding subject. The publications of papers are selected through double peer reviewed to ensure originality, relevance, and readability. The articles published in our journal can be accessed online.
Multiple Choice Questions on JAVA (object oriented programming) bank 3 -- cla...
Multiple Choice Questions on JAVA (object oriented programming) bank 3 -- cla...Kuntal Bhowmick This document contains a 17-page MCQ quiz on object-oriented programming concepts like classes, objects, and conditional statements. It includes 18 multiple choice questions about topics such as the difference between classes and objects, access specifiers, method overloading, constructors, and if statements. Each question is presented on an even page with the corresponding explanation and answer on the adjacent odd page. The document instructs readers to first attempt each question themselves before checking the solution.
Personalized Defect Prediction
Personalized Defect PredictionSung Kim Personalized defect prediction models can more accurately predict buggy changes. The researchers propose two personalized approaches:
1) Personalized Change Classification (PCC) trains a separate model for each developer using their change history.
2) Confidence-based Hybrid PCC (PCC+) combines the predictions from the CC and PCC models, selecting the one with the highest confidence.
The approaches were evaluated on six projects, finding up to 155 more bugs by inspecting only 20% of code locations compared to non-personalized models. PCC and PCC+ consistently outperformed the baseline across different settings, demonstrating the benefits of personalization.
powerpoint feb
powerpoint febimu409 The document proposes using an ensemble of K-nearest neighbor classifiers optimized with genetic programming for intrusion detection. It trains multiple K-NN classifiers on subsets of the KDD Cup 1999 intrusion detection dataset and then uses genetic programming to combine the classifiers to improve performance. Results show the ensemble approach reduces error rates compared to individual classifiers and the genetic programming-based ensemble achieves an area under the ROC curve of 0.99976, outperforming the component classifiers.
Magic behind the numbers - software metrics in practice
Magic behind the numbers - software metrics in practiceSebastian Marek We use static code analysis tools more often these days that create great reports and funky graphs. But do we understand what it all mean?
Software metrics tends to be magic numbers for a lot of people, but they don't really have to be. Let me introduce you to a few basic and
most popular software metric and tools and explain you what they mean and how you can use them to produce better software.
Profiling distributed Java applications
Profiling distributed Java applicationsConstantine Slisenka English:
Sometimes java applications are not as fast as we expect. What if our system contains of hundreds JVMs, databases and other components? We try using profiler, however can't find the bottleneck.
In this talk we discuss:
1. How to profile single JVM
- what is profiler and how does it work
- write simple profiler using java agent and byte-code modification
2. How to profile distributed system
- how engineers in Google doing this
- look at commercial and open source solutions: Dynatrace and Zipkin
- connect to demo-system and see live demos
Russian:
Порой наши java-программы работают медленней, чем хотелось бы. Иногда мы даже подключаемся профайлером, чтобы посмотреть, где тормозит.
А что если наша система состоит из десятков/сотен JVM, баз данных и других компонентов?
На этом техтолке мы обсудим:
1. Как профилировать одну JVM
- что такое профайлер и как он работает под капотом
- напишем простой профайлер с помощью java-агента и байт-код модификаций
2. Как профилировать сложную распределённую систему
- разберёмся как это делают инженеры в Google
- посмотрим готовые коммерческие и open-source решения: Dynatrace и Zipkin
- подключимся к демо-системе и увидим всё своими глазами
https://github.com/kslisenko/java-performance
Design verification--the-past-present-and-future
Design verification--the-past-present-and-futureObsidian Software The document discusses design verification (DV) in the past, present, and future. In the past, DV techniques were basic like visual inspection and code coverage was in its infancy. Now, DV has evolved with more research, texts, and use of techniques like constrained random testing and static verification. The future of DV faces challenges of dealing with increasing capacity needs and being smarter about reaching verification goals. Formal verification and new algorithms are emerging to help address these challenges.
Design Verification: The Past, Present and Futurere
Design Verification: The Past, Present and FuturereDVClub The document discusses design verification past, present and future. It summarizes that in the past DV techniques were basic, the field has since evolved greatly, and the future brings challenges of dealing with increasing capacity and being smarter about verification goals. Formal methods and emulation may help address capacity issues, while new automated approaches could help with coverage closure and stimulus generation.
Scam12.ppt
Scam12.pptYann-Gaël Guéhéneuc This document summarizes a research paper that proposes an approach called LIBCROOS to improve bug location using both information retrieval techniques and binary class relationships. It introduces the problem of bug location and limitations of existing IR techniques. The proposed approach uses IR, binary class relationships like inheritance, aggregation, and association, and a ranker to generate a ranked list of potential buggy classes. Case studies on four open-source projects show LIBCROOS is more accurate than IR alone by decreasing the rank of culprit classes. Inheritance is the most important relationship, followed by aggregation, use, and association.
Object Detection with Transformers
Object Detection with TransformersDatabricks Object detection is a central problem in computer vision and underpins many applications from medical image analysis to autonomous driving. In this talk, we will review the basics of object detection from fundamental concepts to practical techniques. Then, we will dive into cutting-edge methods that use transformers to drastically simplify the object detection pipeline while maintaining predictive performance. Finally, we will show how to train these models at scale using Determined’s integrated deep learning platform and then serve the models using MLflow.
What you will learn:
Basics of object detection including main concepts and techniques
Main ideas from the DETR and Deformable DETR approaches to object detection
Overview of the core capabilities of Determined’s deep learning platform, with a focus on its support for effortless distributed training
How to serve models trained in Determined using MLflow
深度學習在AOI的應用
深度學習在AOI的應用CHENHuiMei This document discusses using fully convolutional neural networks for defect inspection. It begins with an agenda that outlines image segmentation using FCNs and defect inspection. It then provides details on data preparation including labeling guidelines, data augmentation, and model setup using techniques like deconvolution layers and the U-Net architecture. Metrics for evaluating the model like Dice score and IoU are also covered. The document concludes with best practices for successful deep learning projects focusing on aspects like having a large reusable dataset, feasibility of the problem, potential payoff, and fault tolerance.
Predicting Subsystem Defects using Dependency Graph Complexities 
Predicting Subsystem Defects using Dependency Graph Complexities Thomas Zimmermann The document presents a method for predicting subsystem failures in software projects using dependency graph complexities. It finds that code dependency cycles and large cliques (fully connected subgraphs) correlate with more defects. It also finds that measures of complexity in a subsystem's dependency graph, such as the number of nodes and edges, can be used to predict the number of defects in the subsystem. Evaluating the method on a large Microsoft software project showed it could effectively rank subsystems by likelihood of failure.
Rails Software Metrics
Rails Software Metricschiel This document discusses various software metrics that can be used to analyze Ruby on Rails code quality, including lines of code, test coverage, code complexity, and cyclomatic complexity. It provides examples of metrics for two sample Rails applications called Mephisto and Typo and notes that tracking metrics over time is more useful than a single snapshot for understanding code quality issues.
Predicting Defects in SAP Java Code: An Experience Report
Predicting Defects in SAP Java Code: An Experience Reporttilman.holschuh Which components of a large software system are the
most defect-prone? In a study on a large SAP Java system,
we evaluated and compared a number of defect predictors,
based on code features such as complexity metrics, static
error detectors, change frequency, or component imports,
thus replicating a number of earlier case studies in an industrial
context. We found the overall predictive power to
be lower than expected; still, the resulting regression models
successfully predicted 50–60% of the 20% most defectprone
components.
13986149 c-pgming-for-embedded-systems
13986149 c-pgming-for-embedded-systemsPRADEEP The document describes the requirements for an embedded software program that will control a thermostat. The program is required to measure and display the temperature, keep time on a clock, accept time settings to define heating and cooling periods, compare the temperature to the set points for the current period and turn external heating and cooling units on and off as needed. It also needs to accept manual overrides and not change the state of external units too frequently to allow them to operate properly. The hardware requirements include a built-in debug interface to allow for customization and repairs on site.
2010 06-24 karlsruher entwicklertag
2010 06-24 karlsruher entwicklertagMarcel Bruch This document discusses Eclipse code recommenders, which leverage collective intelligence in software development. Eclipse code recommenders observe how frameworks are commonly used and extended to provide intelligent code completion, usage-driven Javadocs, code example recommendations, and smart bug detection. They work by mining development data to extract patterns on how frameworks are reused and then surface those recommendations to developers. The system aims to help developers by recommending what others frequently do based on past usage. It analyzes collected data on framework usage to build recommender models and provide suggestions to developers.
Semantic-Aware Code Model: Elevating the Future of Software Development
Semantic-Aware Code Model: Elevating the Future of Software DevelopmentBaishakhi Ray Keynote: How to build smart models for code generation
Studying the impact of dependency network measures on software quality 
Studying the impact of dependency network measures on software quality ICSM 2010 This document studied the impact of dependency network measures on software quality. It found that incorporating social network analysis (SNA) measures into bug prediction models improved recall and precision by 25% over models using traditional software metrics alone. Specifically, local network measures like cluster fan-in and layer bypass had the most significant impact. The models were evaluated on the Eclipse project and showed potential to prioritize testing efforts and reduce costs.
BSSML17 - Deepnets
BSSML17 - DeepnetsBigML, Inc This document discusses deepnets, which are a type of supervised learning algorithm for classification and regression. Deepnets build upon logistic regression by adding hidden layers between the input and output layers. This allows deepnets to model more complex nonlinear relationships than logistic regression. While deepnets have powerful representational abilities, their success depends on finding the optimal network structure for a given problem. The document outlines how BigML uses metalearning and network search techniques to automate this process and make deepnets more accessible for users. Deepnets work best for problems where computational resources allow exploring many network structures to find the best performing one.
Coding Naked
Coding NakedCaleb Jenkins Caleb Jenkins discusses best practices for writing automated unit tests, including having a test runner, setting the test context or scene, and handling dependencies through techniques like dependency injection and mocking. He advocates writing tests first to define requirements and ensure code meets expectations. Jenkins also addresses challenges with testing edges or interfaces and advocates separating UI/data logic from edges to increase testability.
MARS presentation
MARS presentationAlessio Bockmann The document discusses a tool called MARS that provides software refactoring recommendations based on module coupling and cohesion metrics. It aims to improve software architectures that have drifted from their original design. MARS analyzes external dependencies between modules and internal dependencies within modules to identify candidates for refactoring where a module's coupling is greater than its cohesion. The goal is to recommend moves to reduce coupling and improve cohesion, thereby modulating the software architecture.
CCNA training 101
CCNA training 101Rohan Reddy The document provides information about CCNA training and certification. It discusses the topics covered in the CCNA exam, recommended training courses, study materials, exam format and structure. The CCNA certification tests knowledge of network fundamentals, switching, routing, WAN technologies, security and management. Exams last 90 minutes and contain around 50-60 multiple choice and simulation questions. Common jobs requiring the CCNA include network administrator, database administrator and help desk technician.
Measuring maintainability; software metrics explained
Measuring maintainability; software metrics explainedDennis de Greef In a world of ever-changing business requirements, how can you keep your software moving at the same pace?
If you keep adding lines of code around the previous iteration to add new functionality, things can become complex quite fast.
By measuring complexity, you can resolve and prevent bugs, while measuring class responsibility can make refactoring easier, for example.
In this talk Dennis will go through certain concepts of analysing software with automated tools to spit out numbers which tell a story about your code.
Transcription Factor DNA Binding Prediction
Transcription Factor DNA Binding PredictionUT, San Antonio This project aims to build a binary classifier model to label unlabeled DNA sequences as either positive (p) or negative (n) based on labeled training sequences. The team will take two approaches: 1) A k-mer approach that generates all DNA sequence fragments of length K and counts frequencies to use as attributes for classification models. 2) A PWM approach that uses motif finding tools to generate position weight matrices and score sequences to use as attributes. The approaches will be evaluated individually and combined to obtain the best performing model. Key challenges include deriving meaningful attributes from the sequence data alone. Parameters like k-mer length, number of motifs, and motif lengths will be tuned to optimize model performance.
Software Analytics = Sharing Information
Software Analytics = Sharing InformationThomas Zimmermann This document summarizes a presentation on software analytics. It discusses definitions of software analytics, types of analytics like web and game analytics, the history and various names of software analytics. It outlines key information needs for software development analytics like summarization, alerts, trends and benchmarks. Smart analytics is described as being actionable, real-time, diverse, and sharing insights, methods, models and data. An example is provided of sharing a defect prediction model and discussing challenges with cross-project prediction. Finally, an example is given of sharing insights around skill in the game Halo Reach.
MSR 2013 Preview
MSR 2013 PreviewThomas Zimmermann A preview of the MSR 2013 conference, May 18-19, 2013, in San Francisco, CA. REGISTER NOW! Early registration discounts until April 14. http://msrconf.org
Ad
More Related Content
Similar to Predicting Defects using Network Analysis on Dependency Graphs (20)
Magic behind the numbers - software metrics in practice
Magic behind the numbers - software metrics in practiceSebastian Marek We use static code analysis tools more often these days that create great reports and funky graphs. But do we understand what it all mean?
Software metrics tends to be magic numbers for a lot of people, but they don't really have to be. Let me introduce you to a few basic and
most popular software metric and tools and explain you what they mean and how you can use them to produce better software.
Profiling distributed Java applications
Profiling distributed Java applicationsConstantine Slisenka English:
Sometimes java applications are not as fast as we expect. What if our system contains of hundreds JVMs, databases and other components? We try using profiler, however can't find the bottleneck.
In this talk we discuss:
1. How to profile single JVM
- what is profiler and how does it work
- write simple profiler using java agent and byte-code modification
2. How to profile distributed system
- how engineers in Google doing this
- look at commercial and open source solutions: Dynatrace and Zipkin
- connect to demo-system and see live demos
Russian:
Порой наши java-программы работают медленней, чем хотелось бы. Иногда мы даже подключаемся профайлером, чтобы посмотреть, где тормозит.
А что если наша система состоит из десятков/сотен JVM, баз данных и других компонентов?
На этом техтолке мы обсудим:
1. Как профилировать одну JVM
- что такое профайлер и как он работает под капотом
- напишем простой профайлер с помощью java-агента и байт-код модификаций
2. Как профилировать сложную распределённую систему
- разберёмся как это делают инженеры в Google
- посмотрим готовые коммерческие и open-source решения: Dynatrace и Zipkin
- подключимся к демо-системе и увидим всё своими глазами
https://github.com/kslisenko/java-performance
Design verification--the-past-present-and-future
Design verification--the-past-present-and-futureObsidian Software The document discusses design verification (DV) in the past, present, and future. In the past, DV techniques were basic like visual inspection and code coverage was in its infancy. Now, DV has evolved with more research, texts, and use of techniques like constrained random testing and static verification. The future of DV faces challenges of dealing with increasing capacity needs and being smarter about reaching verification goals. Formal verification and new algorithms are emerging to help address these challenges.
Design Verification: The Past, Present and Futurere
Design Verification: The Past, Present and FuturereDVClub The document discusses design verification past, present and future. It summarizes that in the past DV techniques were basic, the field has since evolved greatly, and the future brings challenges of dealing with increasing capacity and being smarter about verification goals. Formal methods and emulation may help address capacity issues, while new automated approaches could help with coverage closure and stimulus generation.
Scam12.ppt
Scam12.pptYann-Gaël Guéhéneuc This document summarizes a research paper that proposes an approach called LIBCROOS to improve bug location using both information retrieval techniques and binary class relationships. It introduces the problem of bug location and limitations of existing IR techniques. The proposed approach uses IR, binary class relationships like inheritance, aggregation, and association, and a ranker to generate a ranked list of potential buggy classes. Case studies on four open-source projects show LIBCROOS is more accurate than IR alone by decreasing the rank of culprit classes. Inheritance is the most important relationship, followed by aggregation, use, and association.
Object Detection with Transformers
Object Detection with TransformersDatabricks Object detection is a central problem in computer vision and underpins many applications from medical image analysis to autonomous driving. In this talk, we will review the basics of object detection from fundamental concepts to practical techniques. Then, we will dive into cutting-edge methods that use transformers to drastically simplify the object detection pipeline while maintaining predictive performance. Finally, we will show how to train these models at scale using Determined’s integrated deep learning platform and then serve the models using MLflow.
What you will learn:
Basics of object detection including main concepts and techniques
Main ideas from the DETR and Deformable DETR approaches to object detection
Overview of the core capabilities of Determined’s deep learning platform, with a focus on its support for effortless distributed training
How to serve models trained in Determined using MLflow
深度學習在AOI的應用
深度學習在AOI的應用CHENHuiMei This document discusses using fully convolutional neural networks for defect inspection. It begins with an agenda that outlines image segmentation using FCNs and defect inspection. It then provides details on data preparation including labeling guidelines, data augmentation, and model setup using techniques like deconvolution layers and the U-Net architecture. Metrics for evaluating the model like Dice score and IoU are also covered. The document concludes with best practices for successful deep learning projects focusing on aspects like having a large reusable dataset, feasibility of the problem, potential payoff, and fault tolerance.
Predicting Subsystem Defects using Dependency Graph Complexities
Predicting Subsystem Defects using Dependency Graph Complexities Thomas Zimmermann The document presents a method for predicting subsystem failures in software projects using dependency graph complexities. It finds that code dependency cycles and large cliques (fully connected subgraphs) correlate with more defects. It also finds that measures of complexity in a subsystem's dependency graph, such as the number of nodes and edges, can be used to predict the number of defects in the subsystem. Evaluating the method on a large Microsoft software project showed it could effectively rank subsystems by likelihood of failure.
Rails Software Metrics
Rails Software Metricschiel This document discusses various software metrics that can be used to analyze Ruby on Rails code quality, including lines of code, test coverage, code complexity, and cyclomatic complexity. It provides examples of metrics for two sample Rails applications called Mephisto and Typo and notes that tracking metrics over time is more useful than a single snapshot for understanding code quality issues.
Predicting Defects in SAP Java Code: An Experience Report
Predicting Defects in SAP Java Code: An Experience Reporttilman.holschuh Which components of a large software system are the
most defect-prone? In a study on a large SAP Java system,
we evaluated and compared a number of defect predictors,
based on code features such as complexity metrics, static
error detectors, change frequency, or component imports,
thus replicating a number of earlier case studies in an industrial
context. We found the overall predictive power to
be lower than expected; still, the resulting regression models
successfully predicted 50–60% of the 20% most defectprone
components.
13986149 c-pgming-for-embedded-systems
13986149 c-pgming-for-embedded-systemsPRADEEP The document describes the requirements for an embedded software program that will control a thermostat. The program is required to measure and display the temperature, keep time on a clock, accept time settings to define heating and cooling periods, compare the temperature to the set points for the current period and turn external heating and cooling units on and off as needed. It also needs to accept manual overrides and not change the state of external units too frequently to allow them to operate properly. The hardware requirements include a built-in debug interface to allow for customization and repairs on site.
2010 06-24 karlsruher entwicklertag
2010 06-24 karlsruher entwicklertagMarcel Bruch This document discusses Eclipse code recommenders, which leverage collective intelligence in software development. Eclipse code recommenders observe how frameworks are commonly used and extended to provide intelligent code completion, usage-driven Javadocs, code example recommendations, and smart bug detection. They work by mining development data to extract patterns on how frameworks are reused and then surface those recommendations to developers. The system aims to help developers by recommending what others frequently do based on past usage. It analyzes collected data on framework usage to build recommender models and provide suggestions to developers.
Semantic-Aware Code Model: Elevating the Future of Software Development
Semantic-Aware Code Model: Elevating the Future of Software DevelopmentBaishakhi Ray Keynote: How to build smart models for code generation
Studying the impact of dependency network measures on software quality
Studying the impact of dependency network measures on software quality ICSM 2010 This document studied the impact of dependency network measures on software quality. It found that incorporating social network analysis (SNA) measures into bug prediction models improved recall and precision by 25% over models using traditional software metrics alone. Specifically, local network measures like cluster fan-in and layer bypass had the most significant impact. The models were evaluated on the Eclipse project and showed potential to prioritize testing efforts and reduce costs.
BSSML17 - Deepnets
BSSML17 - DeepnetsBigML, Inc This document discusses deepnets, which are a type of supervised learning algorithm for classification and regression. Deepnets build upon logistic regression by adding hidden layers between the input and output layers. This allows deepnets to model more complex nonlinear relationships than logistic regression. While deepnets have powerful representational abilities, their success depends on finding the optimal network structure for a given problem. The document outlines how BigML uses metalearning and network search techniques to automate this process and make deepnets more accessible for users. Deepnets work best for problems where computational resources allow exploring many network structures to find the best performing one.
Coding Naked
Coding NakedCaleb Jenkins Caleb Jenkins discusses best practices for writing automated unit tests, including having a test runner, setting the test context or scene, and handling dependencies through techniques like dependency injection and mocking. He advocates writing tests first to define requirements and ensure code meets expectations. Jenkins also addresses challenges with testing edges or interfaces and advocates separating UI/data logic from edges to increase testability.
MARS presentation
MARS presentationAlessio Bockmann The document discusses a tool called MARS that provides software refactoring recommendations based on module coupling and cohesion metrics. It aims to improve software architectures that have drifted from their original design. MARS analyzes external dependencies between modules and internal dependencies within modules to identify candidates for refactoring where a module's coupling is greater than its cohesion. The goal is to recommend moves to reduce coupling and improve cohesion, thereby modulating the software architecture.
CCNA training 101
CCNA training 101Rohan Reddy The document provides information about CCNA training and certification. It discusses the topics covered in the CCNA exam, recommended training courses, study materials, exam format and structure. The CCNA certification tests knowledge of network fundamentals, switching, routing, WAN technologies, security and management. Exams last 90 minutes and contain around 50-60 multiple choice and simulation questions. Common jobs requiring the CCNA include network administrator, database administrator and help desk technician.
Measuring maintainability; software metrics explained
Measuring maintainability; software metrics explainedDennis de Greef In a world of ever-changing business requirements, how can you keep your software moving at the same pace?
If you keep adding lines of code around the previous iteration to add new functionality, things can become complex quite fast.
By measuring complexity, you can resolve and prevent bugs, while measuring class responsibility can make refactoring easier, for example.
In this talk Dennis will go through certain concepts of analysing software with automated tools to spit out numbers which tell a story about your code.
Transcription Factor DNA Binding Prediction
Transcription Factor DNA Binding PredictionUT, San Antonio This project aims to build a binary classifier model to label unlabeled DNA sequences as either positive (p) or negative (n) based on labeled training sequences. The team will take two approaches: 1) A k-mer approach that generates all DNA sequence fragments of length K and counts frequencies to use as attributes for classification models. 2) A PWM approach that uses motif finding tools to generate position weight matrices and score sequences to use as attributes. The approaches will be evaluated individually and combined to obtain the best performing model. Key challenges include deriving meaningful attributes from the sequence data alone. Parameters like k-mer length, number of motifs, and motif lengths will be tuned to optimize model performance.
More from Thomas Zimmermann (20)
Software Analytics = Sharing Information
Software Analytics = Sharing InformationThomas Zimmermann This document summarizes a presentation on software analytics. It discusses definitions of software analytics, types of analytics like web and game analytics, the history and various names of software analytics. It outlines key information needs for software development analytics like summarization, alerts, trends and benchmarks. Smart analytics is described as being actionable, real-time, diverse, and sharing insights, methods, models and data. An example is provided of sharing a defect prediction model and discussing challenges with cross-project prediction. Finally, an example is given of sharing insights around skill in the game Halo Reach.
MSR 2013 Preview
MSR 2013 PreviewThomas Zimmermann A preview of the MSR 2013 conference, May 18-19, 2013, in San Francisco, CA. REGISTER NOW! Early registration discounts until April 14. http://msrconf.org
Predicting Method Crashes with Bytecode Operations
Predicting Method Crashes with Bytecode OperationsThomas Zimmermann 1) The document presents a technique called ReCrash+ that combines crash prediction using bytecode features with capture and replay to reproduce crashes with lower overhead.
2) It evaluates crash prediction both within and across projects, finding that bytecode features can accurately identify crashed methods.
3) An experiment applying ReCrash+ to a project called SVNKit was able to reproduce 3 crashes by only monitoring the small subset of methods predicted as crash-prone.
Analytics for smarter software development 
Analytics for smarter software development Thomas Zimmermann The document discusses how software analytics can empower software development teams to make better decisions by gaining insight from data. It describes how analytics involves using analysis, data, and systematic reasoning to make decisions. The document argues that software analytics can help teams share insights from their data and empower them to gain a better understanding of their software development process.
Characterizing and Predicting Which Bugs Get Reopened
Characterizing and Predicting Which Bugs Get ReopenedThomas Zimmermann This document summarizes a study characterizing and predicting which software bugs get reopened. Through a qualitative survey, researchers identified six main causes of bug reopens: bugs being difficult to reproduce, misunderstood root causes, insufficient bug information, increased bug priority later on, regression bugs, and process-related issues. A quantitative analysis using logistic regression found that bugs found through code analysis tools or human review were less likely to reopen, while bugs found through system or customer testing were more likely to reopen. The analysis also found that bugs opened by people with higher reputations based on past bugs were less likely to reopen.
Data driven games user research
Data driven games user researchThomas Zimmermann This document discusses Microsoft's use of data-driven user research methods for game development. It summarizes different types of data collected, including in-game telemetry, Xbox Live usage data, and online forum discussions. Examples are provided of analyses conducted on data from games like Project Gotham Racing 4 and Halo 3 to understand player behavior and identify underutilized game elements. The goal is to better understand the player experience and Xbox ecosystem to inform game design and personalization.
Not my bug! Reasons for software bug report reassignments
Not my bug! Reasons for software bug report reassignmentsThomas Zimmermann This document discusses reasons for software bug report reassignments in Microsoft products. Through qualitative surveys and analysis of bug reports for Windows Vista, it identifies the top reasons for reassignments as determining the root cause, unclear ownership between teams, poor bug report quality, determining the proper fix, and workload balancing. It also finds that bug reports with reassignment cycles at the beginning are more likely to be fixed, while cycles at the end indicate risk of the bug not being addressed. The lessons learned point to making the bug fixing process more collaborative, fluid and coordinated.
Empirical Software Engineering at Microsoft Research
Empirical Software Engineering at Microsoft ResearchThomas Zimmermann An invited talk that I gave in Tokyo. Very special thanks to Shuji Morisaki who was my translator during the session. Many thanks to Chris Bird, Nachi Nagappan, Rahul Premraj, and Sascha Just who provided slides for this talk.
Security trend analysis with CVE topic models
Security trend analysis with CVE topic modelsThomas Zimmermann This document discusses using topic models to analyze trends in vulnerability data from Common Vulnerabilities and Exposures (CVE). Latent Dirichlet allocation is used to identify topics in CVE summaries from 2000-2009. Key trends identified include a decline in buffer overflows and format strings but an increase in SQL injection, cross-site scripting, and application servers. The topics identified generally align well with the Common Weakness Enumeration classifications.
Analytics for software development
Analytics for software developmentThomas Zimmermann 1) The document discusses software analytics and how it can be used to understand what happened in the past, what is currently happening, and what may happen in the future for software development projects.
2) It notes that different stakeholders like researchers, developers, testers, and managers have different needs for analytics and a single tool may not be enough to meet all needs.
3) The presentation outlines challenges with software analytics including data collection, data quality, privacy, understanding different user needs, and user experience. It stresses the importance of making data actionable and accessible for stakeholders.
Characterizing and predicting which bugs get fixed
Characterizing and predicting which bugs get fixedThomas Zimmermann This document discusses factors that affect whether software bugs get fixed. It analyzes bug report data from Microsoft and surveys Microsoft employees. The analysis found that bugs with one reassignment are more likely to be fixed, but too many reassignments decrease likelihood. Bugs opened by people with high reputations and handled within single teams and locations are more likely to be fixed. The survey revealed social factors like seniority, reputation, relationships, and trust also impact the bug fixing process.
Cross-project defect prediction
Cross-project defect predictionThomas Zimmermann This document examines cross-project defect prediction, where a model trained on one project is used to predict defects in another project. In an experiment of 622 cross-project combinations across 12 systems, only 3.4% had successful predictions. However, certain project similarities like domain and characteristics like code reviews increased prediction precision and recall. Decision trees were created to help select project combinations where cross-project prediction is more likely to succeed based on characteristics like intended audience and pre-release bug metrics.
Changes and Bugs: Mining and Predicting Development Activities
Changes and Bugs: Mining and Predicting Development ActivitiesThomas Zimmermann 1. The document discusses mining software archives and predicting development activities through analyzing version control data, bug databases, and code dependencies.
2. Key contributions include fine-grained analysis of version archives to identify patterns and cross-cutting changes, and mining bug databases to predict defects based on dependencies and historical data.
3. Future work involves using social networking data and collaboration information to improve defect prediction.
Quality of Bug Reports in Open Source
Quality of Bug Reports in Open SourceThomas Zimmermann The document discusses research into what makes a good bug report based on a survey of over 150 developers. It finds that the most helpful items for fixing bugs are steps to reproduce, stack traces, and observed behavior. The biggest problems causing delays are incomplete information, wrong steps to reproduce, and wrong expected behavior. The research also measured bug report quality and found stack traces and readability correlated with shorter fix times.
Meet Tom and his Fish
Meet Tom and his FishThomas Zimmermann The document discusses software mining research including predicting bugs, defects, and vulnerabilities. It examines usage patterns and recommends systems to address cross-cutting concerns. The research analyzes past software projects and aims to guide future development.
Got Myth? Myths in Software Engineering
Got Myth? Myths in Software EngineeringThomas Zimmermann The document summarizes a working session on software engineering myths held on October 4, 2007 in Paris. The session included discussions of common myths in areas such as complexity, bugs, clones, aspect-oriented programming, and project management. Participants shared and debated myths in small group discussions. Some examples of myths discussed include the ideas that complexity leads to more bugs, clones are always bad, and aspect-oriented programs are inherently easier to maintain. The session aimed to evaluate whether such perceptions are truly myths or if aspects of them have been found to hold true in various studies.
Mining Workspace Updates in CVS
Mining Workspace Updates in CVSThomas Zimmermann The document discusses updates and commits made in CVS version control software. It includes logs of revisions for an IBuffer.java file with information on dates, authors, and changes made. It also contains charts showing data on how often updates integrate changes, how many commits lead to integrations, what percentage of integrations have conflicts, and how many conflicting changes are discarded for different projects like JEdit and Python. The conclusion is that CVS can integrate many changes but not all, and developers often have their own workspaces.
Mining Software Archives to Support Software Development
Mining Software Archives to Support Software DevelopmentThomas Zimmermann 1. The document discusses mining software archives and repositories to help guide software developers and predict defects.
2. It describes the eROSE tool which mines past associations between code changes to suggest related files and locations to developers.
3. The BugCache model predicts future defects based on the hypothesis that defects are temporally local, with a cache that loads elements likely to have defects.
Unit testing with JUnit
Unit testing with JUnitThomas Zimmermann Unit testing with JUnit focuses on testing individual units of source code using JUnit. Key aspects covered include:
- Unit testing tests the smallest testable parts of code, while integration testing tests groups of code modules and system testing tests complete integrated systems.
- Tests in JUnit are implemented as public void testX() methods that call methods under test and assert expected outcomes.
- A test suite assembles individual test methods using reflection and executes the suite to automatically run all tests.
- Best practices include writing tests before code, testing all code that could break, and running tests frequently for rapid feedback.
Ad
Recently uploaded (20)
The GensoL Scam: How 1,000 Cr EV Dreams Funded Jaggis' Luxury Empire
The GensoL Scam: How 1,000 Cr EV Dreams Funded Jaggis' Luxury EmpireCACompany Gensol Engineering secured ₹829 crore to build electric vehicles but diverted ₹262 crore to luxury purchases, including apartments and Rolexes. SEBI exposed the fraud, banned the founders, and the company's stock crashed 91%, wiping out investor wealth. This scandal highlights critical gaps in corporate governance and startup accountability in India's EV sector.
Dimensions of Globalisation .pptx.Globalization is a multifaceted process wit...
Dimensions of Globalisation .pptx.Globalization is a multifaceted process wit...Vinodha Devi Globalization is a multifaceted process with several interconnected dimensions. Here are the key dimensions of globalization:
The Role of Storytelling in Legal Communication (www.kiu.ac.ug)
The Role of Storytelling in Legal Communication (www.kiu.ac.ug)publication11 Storytelling is a powerful tool in legal communication, serving as a means of making complex legal
arguments more comprehensible and persuasive. This paper investigates how storytelling enhances
advocacy by engaging the emotions and empathy of judges, jurors, and other legal audiences. It delves
into the significance of narrative techniques in trial proceedings, legal writing, and negotiation,
examining how storytelling can simplify intricate legal matters while maintaining accuracy and ethical
integrity. The study further discusses strategies for integrating storytelling into legal practice,
emphasizing the selection of appropriate narratives and their impact on legal decision-making. Through
case studies and theoretical analysis, this paper highlights how effective storytelling contributes to
successful legal advocacy while addressing the ethical considerations surrounding its use.
META Stock: Meta Platforms' Bold Bet on the Future
META Stock: Meta Platforms' Bold Bet on the FutureMeyka ai META stock represents Meta Platforms, Inc., a tech giant known for Facebook, Instagram, and WhatsApp. With strong advertising revenue, major investments in AI, and long-term plans for the metaverse, META stock offers growth potential for investors focused on innovation and digital trends
TriStar Gold Corporate Presentation May 2025
TriStar Gold Corporate Presentation May 2025Adnet Communications TriStar Gold Corporate Presentation May 2025
Silver One May 2025 Corporate Presentation
Silver One May 2025 Corporate PresentationAdnet Communications Silver One May 2025 Corporate Presentation
Understanding Cryptocurrency Guide For Beginners (2025).pdf
Understanding Cryptocurrency Guide For Beginners (2025).pdfSERP Navigator Curious about Bitcoin, Ethereum, and the future of digital finance? Understanding Crypto is your essential beginner’s guide to the world of cryptocurrency. This easy-to-follow presentation breaks down the core concepts of crypto—including what it is, how it works, the different types of cryptocurrencies, and the revolutionary technology behind it: blockchain.
You’ll also explore key topics like mining, wallets, exchanges, real-world use cases, investor risks, legal considerations, and the future of decentralized finance. Whether you’re new to crypto or simply looking to sharpen your knowledge, this guide delivers practical insights in a clear, structured way.
Created with educational support from resources like Cryptonary.com, this presentation empowers you to navigate the crypto space confidently and responsibly. Dive in and start your crypto journey today.
Hard Money Loan – Fast Cash, But at What Cost.pptx
Hard Money Loan – Fast Cash, But at What Cost.pptxameliajena8 Need fast funding for your next real estate deal? A hard money loan might seem like the perfect solution—quick approvals, no credit checks, and money in hand fast. But beware: the dark side includes high interest rates, short terms, and even predatory practices. Know the red flags before you sign.
Ad