How to translate ML predictions to practical solutions, by Kaggle Learn Creator

Our guest speaker Dan Becker is a founder of Kaggle Learn and the CEO of Decision.AI. Singularly gifted at blending high level mathematical and pragmatic thinking, Dan plays a vital role the growth of Data Science and Machine Learning (ML) worldwide.

Title: How to translate ML predictions into practically useful solutions

Abstract: Machine learning is fun, impressive and gets a lot of attention. But most models are never put into production. Why do so many projects produce no practical value?

I see the biggest problem as a missing step in conventional ML workflows. The API to conventional ML models is a single predict function. But real-world solutions require another step to go from prediction to practical decisions.

For example, if a model predicts a transaction is 1% likely to be fraudulent, should you reject it? If a model predicts a restaurant will sell 100 salmon in the next week, how many should they purchase from their wholesaler to ensure they don't disappoint customers?

In this talk, I'll explain key ideas from the field of decision theory to help you turn ML predictions into actionable and value producing solutions.

Speaker Bio: Dan Becker is a Data Scientist with a PhD in econometrics. He finished in 2nd place (out of 1353 teams) in the Heritage Health Prize machine learning competition, and he has contributed to open source libraries deep learning including Keras and PyLearn2.

Dan created Kaggle Learn, a free AI education platform used by 45,000 people per month. Dan is the founder of Decision AI, which is applying insights from decision theory to make machine learning models more practically useful.

The event is organized by the Bugout.dev team who builds the search engine for programmers. Join our Slack community where programmers are helping programmers: https://join.slack.com/t/bugout-dev/shared_invite/zt-fhepyt87-5XcJLy0iu702SO_hMFKNhQ