Categorical inputs
Supervised Learning in R: Regression
Nina Zumel and John Mount
Win-Vector, LLC
Example: Effect of Diet on Weight Loss
WtLoss24 ~ Diet + Age + BMI
| Diet | Age | BMI | WtLoss24 |
|---|---|---|---|
| Med | 59 | 30.67 | -6.7 |
| Low-Carb | 48 | 29.59 | 8.4 |
| Low-Fat | 52 | 32.9 | 6.3 |
| Med | 53 | 28.92 | 8.3 |
| Low-Fat | 47 | 30.20 | 6.3 |
model.matrix()
model.matrix(WtLoss24 ~ Diet + Age + BMI, data = diet)
- All numerical values
- Converts categorical variable with N levels into N - 1 indicator variables
Indicator Variables to Represent Categories
Original Data
| Diet | Age | ... |
|---|---|---|
| Med | 59 | ... |
| Low-Carb | 48 | ... |
| Low-Fat | 52 | ... |
| Med | 53 | ... |
| Low-Fat | 47 | ... |
Model Matrix
| (Int) | DietLow-Fat | DietMed | ... |
|---|---|---|---|
| 1 | 0 | 1 | ... |
| 1 | 0 | 0 | ... |
| 1 | 1 | 0 | ... |
| 1 | 0 | 1 | ... |
| 1 | 1 | 0 | ... |
- reference level: "Low-Carb"
Interpreting the Indicator Variables
Linear Model:
$$ WtLoss24 = \beta_{0} + \beta_{DietLow} x_{DietLow} + \beta_{DietMed} x_{DietMed} + \beta_{Age} x_{Age} + \beta_{BMI} x_{BMI} $$
lm(WtLoss24 ~ Diet + Age + BMI, data = diet))
Coefficients:
(Intercept) DietLow-Fat DietMed
-1.37149 -2.32130 -0.97883
Age BMI
0.12648 0.01262
Issues with one-hot-encoding
- Too many levels can be a problem
- Example: ZIP code (about 40,000 codes)
- Don't hash with geometric methods!
Let's practice!
Supervised Learning in R: Regression
