This is an old revision of the document!
Forecasting Team Meeting - September 121, 2020
Attended: Josh, Keola
Updates
- Completed all regression models
- Topics learned: SVR, Decision Tree Regression (DTR), Random Forest Regression (RFR)
Problems
- COVID-19 forced today's lab hours to be online
Reminders
- 'Fit' typically normalizes data between -3 and 3 and is used when there is a large discrepancy between two sets of data (i.e. x = (1:10), y = (30:100))
- When normalizing data, create new variables (i.e. x_norm) instead of updating original variables
- DTR and RFR are typically used with datasets that have more than one feature
- With DTR, splits are determined through an algorithm while RFR determines splits randomly
- When predicting for values in a range/split, the average value of the independent variable (within that specific split) is taken.