Meeting 3: ML Researcher
Area of Expertise: (Atrial Fibrillation/Heart Failure) + Framingham Heart Study Dataset (Conducted Research since 1988)

Main Takeaways:

​

• Some of the columns are empty or very low with regard to the surrounding values. An imputer is a way to generate missing values.

• You should go through each of the integer values and sort through them to determine whether or not they’re continuous or discrete (Alternatively Categorical).

• Normalize the data. Recommended against dimensionality reduction because it doesn’t “play well” with categorical data.

• YoI could probably run 10 models per day in terms of runtime. (Ex. Support Vector Machines, Dense Neural Networks, Decision Tree)

​

​