Machine Learning is used applied when the scale of data is too large for rule-based systems to handle. For example, you might be able to manually predict purchasing pattern at a small roadside shop by going through the sales entries but it would be difficult to do the same for a large departmental store simply because the number of factors involved are very high.

Before we understand the approach to testing Machine Learning applications, let’s look at the steps involved in building them.

In the previous post, we gave an overview of how ZUJYA helps in validating predictive models. While ZUJYA is built to validate machine learning based predictive models, it can also be used along with standard automation frameworks that are used for Functional/Non-Functional Testing.

The often stated challenges in considering non-functional tests in any model (let alone continuous testing) are: Non-functional tests are inconsistently defined and poorly planned Non-functional tests are often treated as lower priority Lack of suitable skillset

During the past few years majority of the software projects use Continuous Delivery, a software engineering approach in which teams produce software in short cycles, ensuring that software can be reliably released at any time.

Non-functional testing is critical to ensuring any software's full functionality & to ensuring seamless business operations. Here's why...

It was 8 AM during one of the critical release regression times, our SDET came to office and found that 60% of his tests failed due to changes in UI (element IDs changed, new elements introduced, elements removed etc). It is a suite of about 2000 Tests, can we analyse all of these 1200 failed tests in the next few hours and find the reasons?

Those of you who happened to be at the Moscone center in San Francisco this year for the Google I/O conference know would have stood witness to Google CEO Sundar Pichai’s talk about Machine Learning, Artificial Intelligence,