In-sample performance
It's very important to know whether your regression model is useful or not. A useful model can be one that captures the structure of your training set well. One way to assess this in-sample performance is to predict on training data and calculate the mean absolute error of all predicted data points.
In this exercise, you will evaluate your in-sample predictions using MAE (mean absolute error). MAE tells you approximately how far away the predictions are from the true values.
It is calculated using the following formula, where \(n\) is the number of predictions made:
$$MAE = \frac{1}{n} \cdot \sum_{i=1}^n \text{absolute value of the }i\text{th error}$$
Available in your workspace is your model
, the regression tree that you built in the last exercises.
This exercise is part of the course
Machine Learning with Tree-Based Models in R
Exercise instructions
- Create
in_sample_predictions
by usingmodel
to predict on thechocolate_train
tibble. - Calculate a vector
abs_diffs
that contains the absolute differences between the in-sample-predictions and the true grades. - Calculate the mean absolute error according to the formula above.
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
# Predict using the training set
in_sample_predictions <- predict(model,
___)
# Calculate the vector of absolute differences
abs_diffs <- ___(__$___ - ___$___)
# Calculate the mean absolute error
1 / ___ * ___