Kaggle/Time Series
- Time Step feature: derived directly from the time index
- Time dummy: time-step feature counting off steps in the series
- Model time dependence
sklearn.linear_model.LinearRegression().fit(X, y)
- Time dummy: time-step feature counting off steps in the series
- Lag features lag target series by n-steps to make it appear as if they happened later.
- Allows for fitting lag plots where each observation is plotted against the previous one.
df.shift(1)in pandas to offset the values.
- Allows for fitting lag plots where each observation is plotted against the previous one.