1.4 Feature engineering for numbers and categories

Key points

1. One-hot encoding turns a categorical column into one true/false column per category. Categorical means the values are labels, not numbers.

   What NVIDIA says (1)

   “cudf . get_dummies ( df ) b a_value1 a_value2 0 0 True False 1 0 False True 2 0 False False”

   — cuDF API: get_dummies

2. Target encoding replaces a category with a statistic of the label, usually its mean. cuML's TargetEncoder uses folds to avoid label leakage, which is when the label sneaks into a feature.

   What NVIDIA says (2)

   “The input data is grouped by the columns Xs and the aggregated mean value of Y of each group is calculated to replace each value of Xs .”

   — cuML API: TargetEncoder

   “Several optimizations are applied to prevent label leakage and parallelize the execution.”

   — cuML API: TargetEncoder

3. Scaling puts numeric features on a similar range so no feature dominates by its units. StandardScaler computes z = (x - mean) / std and reuses those training statistics later.

   What NVIDIA says (2)

   “Standardize features by removing the mean and scaling to unit variance”

   — cuML API: StandardScaler

   “Mean and standard deviation are then stored to be used on later data using transform() .”

   — cuML API: StandardScaler

4. Binning, also called discretization, groups continuous numbers into intervals. Each bin can then be treated as a category.

   What NVIDIA says (1)

   “Bin continuous data into intervals.”

   — cuML API: KBinsDiscretizer

5. With all categories present, one column can be computed from the others. Dropping one removes that redundancy but changes how a penalized model treats categories.

   What NVIDIA says (1)

   “However, dropping one category breaks the symmetry of the original representation and can therefore induce a bias in downstream models, for instance for penalized linear classification or regression models.”

   — cuML API: OneHotEncoder

Key terms

• Feature: An input column a model learns from.
• Feature engineering: Creating or reshaping input columns so a model can learn better.
• One-hot encoding: Turning a category column into one 0/1 column per category.
• Target encoding: Replacing each category with the average target value for that category.
• Standard scaling: Rescaling a numeric feature to mean 0 and standard deviation 1.
• Binning: Grouping a continuous number into ranges, such as age bands.

Try it

• Feature lab: standard scaling, min-max scaling and one-hot encoding

Sample question

A 'color' column has values red, green and blue. Which cuDF call turns it into one indicator column per value?