Using Machine Learning to Find and Eliminate Sloppy Work in Crowdsourced Testing
At Rainforest, we use crowdsourced QA testers to help our customers do faster manual testing. But relying on a team of thousands of testers around the world can open up potential problems with quality control. Ensuring that the quality of our crowdsourced testing workforce stays high is a critical part of our work at Rainforest QA.
Rainforest QA Data Scientist Maciej Gryka recently spoke at PyData Amsterdam about the challenges that we’ve faced in building a team of qualified Rainforest testers. He discussed how we’ve implemented a combination of better training and incentives, redundancy, and input pattern detection to ensure that Rainforest testers perform consistently and our customers see reliable test results.
Watch the video below to learn how Maciej and his team developed a machine learning model trained on data collected from Rainforest testers to catch sloppy crowdsourced testing work more efficiently.Filed under: machine learning, conferences, and crowdsourced testing