Models that predict: from the line to the trees
Your copilot fits a model in seconds and hands you some coefficients, an error rate and a chart. The hard part is no longer fitting it: it is knowing how to read what is inside. This course goes through, in four blocks, the models most used to predict a quantity or a yes-or-no, and for each one first teaches you to read it: what each coefficient says, what it assumes and which variable might be missing.
Intermediate
Data, Analytics and AI
-
-
Preview
-
-
Block 1: Several variables: reading the knobs of a regression3Lessons · 1 hr 15 min
-
From one knob to several
-
Categories, interaction effects and residuals
-
Block 1 exam10 xp
-
-
Block 2: Forecasting over time: trend, seasonality and the rule without a model3Lessons · 1 hr 15 min
-
Anatomy of a series
-
Forecasting without cheating
-
Block 2 exam10 xp
-
-
Block 3: Classifying: from probability to the tree3Lessons · 1 hr 20 min
-
Predicting a yes or a no
-
A tree: chained questions
-
Block 3 exam10 xp
-
-
Block 4: Many trees, and when not to train anything3Lessons · 1 hr 15 min
-
Forests: many models are better than one
-
Train a model or ask one that is already trained?
-
Block 4 exam10 xp
-