Machine Learning

Simple Regression modeling have failed a billion times to deal with many real world data that oftentimes violates its model assumptions: multi-normal, homoscedasticity, variable independence... With the help of technology boost, especially the computing speed, a new set of tools were able to play important roles in data modeling. Machine Learning has been a must-to-know skill for any data scientist. Thankful to first-class programmers, many great computing tools have been developed to emancipate data scientists from spending long time figuring out implementation, which allowed them to pay more attention to DATA itself.

The hardest part is no longer how to implement the model, but rather how to find the best model that is able to explain/predict the data. Many data scientists stop when they have tried all possible models, while a real experienced data scientist knows what to do next or at the very least knows the reasoning behind various model performance.