Machine Learning for QA
Machine Learning is picking up pace in many industries for multiple purposes. Whether it be self-driving cars, sentiment analysis using natural language processing, making predictions using regression models, or making classifications, its uses and benefits are not to be ignored. Hourglass Software is leveraging Machine Learning for QA which is an unexplored territory. Machine Learning has its own merits in other industries and can also be applied to Quality Assurance as well. Thus, in one of the many areas Machine Learning can be applied to QA, we have used a regression model to predict the probability of defects for a given timeframe in the software development life cycle for specific components of the unit under test.
What is Hourglass Bug Predictor
Hourglass Bug Predictor is a Command Line Tool that integrates with Jira and uses Machine Learning to predict the future quality of your products in development. It will predict how many bugs you should expect in the next seven days, from the time the software is run. It will query your Jira tracking system for Bugs under a specified Jira “Project”. Based on the data it finds in your Jira Management System (Atlassian), it will train a Machine Learning Model and the predict the number of bugs you should expect to find in the following week of your development. For example, it will tell you/predict that there will be 5 bugs in the next week or some other prediction value. The prediction takes into account the fields of your Jira Bugs both for training as well as prediction. The prediction feature vector also takes historical timelines and grouping of your bug data and respective attributes.
An example output would be: Bug Prediction Count for the next 7 days for input parameters:0.5
How Hourglass Bug Predictor Leverages Machine Learning
Hourglass Bug Predictor, which is training the Machine Learning Model using data from your Jira project, and then utilizing a Linear Regression algorithm for prediction. Bug Predictor uses a complex proprietary algorithm to create the training data for the feature vector. It leverages components and fields of you Jira bugs, and other time and count-based groupings, assortments, and calculations. Once the model is trained with fine-tuned feature vector data, the predictions will yield result values with high confidence.
Optimize your QA Resources and Be More Cost-effective
This tool can help optimize your business and make your QA team more efficient. It will tell you which specific areas of your product in development there is a probability of defects. Thus, your QA team can focus more on testing those components of the product and use the predictions to prioritize testing those areas first. This also allows finding defects earlier in the development cycle, which is more cost-effective, and therefore optimizing your business’ cost for development and testing.
To use the tool, ensure your Jira project has enough historical bug information for more accurate predictions. Then, while in your planning stage, run the tool with the input parameters of “where” to make the prediction, which will then predict the potential defects in that area for the next seven days. You can then make smart decisions for where to apply your available QA resources.
Hourglass Software Additional Services
Additionally, Hourglass Software provides QA and Test Automation services that you can out-source. We have a veteran staff that have worked at Amazon as Senior QA, Motorola Solutions, and other companies. They are proficient in various technologies and automation frameworks which you can find here:
Hourglass Bug Predictor Guide and Pricing