In this talk I will discuss the main steps involved in having machine learning models in production (feature engineering, training, tuning and evaluating models, deploying to a scoring engine, evaluating and monitoring the live scoring and the effect of the business actions taken based on the models). Across the talk I will focus on various pitfalls and I will also point out several falsehoods created by the hype around machine learning, “big data” and “artificial intelligence” in recent years.
Chief Data Scientist, Epoch
Szilard has trained, deployed and maintained machine learning models in production (starting with random forests) since 2007. He’s the Chief Scientist at a credit card processing company in Santa Monica and a Physics PhD with 20+ years experience with data, computing and modeling. He founded the first data meetup in Los Angeles in 2009 (by chance) and he has organized several data science events since. He’s also teaching data science and machine learning at CEU in Hungary and at UCLA in California.