Why are we training ML models wrong and how can feature stores help? - Episode 101 - a podcast by SD Times

In this week's episode, we talk about the problem of data leakage, which occurs when data scientists feed data that did not exist during the time of a past event to machine learning models. 

Monte Zweben, CEO of Splice Machine talks about how feature stores can help with this issue by validating when a data set actually occurred and then correcting these point-in-time consistency issues.

Further episodes of What the Dev?

Further podcasts by SD Times

Website of SD Times