Artificial Intelligence and a Versatile Metrics Engine Help Create Better Software
Predictive 3.0 is a powerful artificial intelligence system for managing the quality of your software code throughout the entire life cycle. Predictive 3.0 predicts areas of greatest risk in your C++, Java, C# or C code, so you can prioritize and allocate resources to fix errors now. Save $1000 in maintenance costs for each 1000 lines of source code. With three different AI learners and a versatile metrics engine that allows you to create your own metrics and set your own thresholds, you have the flexibility to use the approach that is most meaningful for your own software project. Use Predictive 3.0 to:
- Monitor and Enforce Quality: Software Leads and QA Professionals can monitor the quality of the code and conduct quality enforcement over the life of a project and across projects, at any point in time. Use the data to support your code reviews or analyze the data after each build. Predictive 3.0 provides four categories of metrics that can be used to evaluate code quality: complexity, operational, quality and user-defined metrics. You can also set and enforce your own thresholds for the project.
- Prioritize Testing: Predictive 3.0 is used by Test or QA Leads to help prioritize testing. The three modes of Predictive 3.0 include QA mode, Heuristics mode (with 3 AI analyzers) and Error Trend mode. Used together, they can help find chronic problems in the code and where they will occur in the future. This enables the team to prioritize and allocate resources to fix the errors before the code is released.
- Refactor and Investigate Source Code: Software Leads use Predictive 3.0 during the maintenance phase to identify areas of code that need to be refactored, finding areas of poor quality, complexity and chronic errors that can result in more errors. The user can set their own thresholds or create their own metric in order to zero in on the areas of risk and fix errors. Developers or Analysts can investigate issues within the source code that might be causing errors. For instance, the user can create a metric to identify a string that may be part of a larger problem and can then task Predictive 3.0 to show all error-related occurrences of the string.
How Predictive 3.0 Works
Predictive 3.0 got its start from NASA, where it was used on large projects to assist in independent validation and verification of code quality and integrity. In this work, it was initially shown that metrics could be used to predict errors in the source code. Using metrics alone to predict errors has inherent instability, but if artificial intelligence is used to analyze the metrics and set thresholds within the project, it becomes more precise. Predictive 3.0 can generate source code metrics three different ways: by using historical NASA data, by applying one of the three AI learners, or as determined by you.
Predictive 3.0 Provides Measurable Value
Return on the investment in software tools can be measured in many ways: time saved, productivity gains, costs eliminated, resources replaced, or increased revenue. The figure below shows three different scenarios for measuring the ROI for Predictive 3.0:
- Software errors can be very costly to fix after software is released; it's estimated that every error found and fixed early in the cycle saves about $500.
- For a project of 500,000 lines of code, Predictive 3.0 will increase the ability to find errors by 5%, which equates to a savings of $200,000 per project.
- Releasing software late to the market can mean a day for day loss in company revenues. Releasing an inferior quality product can damage reputation in the long run, which can cost a company millions or billions. Using Predictive 3.0 can help companies avoid these scenarios, so they can release better quality software, right on time.
