Data Preprocessing Cheat Sheet

This cheat sheet covers the essential steps for preparing raw data for machine learning models, focusing on loading, handling missing values, scaling, and encoding.

Core Principles

Data quality is paramount for model performance.
Preprocessing reduces noise and improves accuracy.
Different models require different data formats.

Action Steps

1. Load your dataset (e.g., Iris, Titanic).
2. Identify and handle missing values (imputation or removal).
3. Normalize or standardize numerical features.
4. Encode categorical features into numerical representations.

Key Terms

Normalization: Scaling features to a fixed range, usually 0 to 1.
Standardization: Scaling features to have zero mean and unit variance.
Imputation: Replacing missing values with substituted values (mean, median, mode).
One-Hot Encoding: Creating new binary columns for each category.
Label Encoding: Assigning a unique integer to each category.

Real World Examples

Titanic Dataset Survival Prediction: Handle missing 'Age' and 'Cabin' values, encode 'Sex' and 'Embarked' features.
Iris Flower Classification: No missing values, but scaling 'SepalLength', 'SepalWidth', 'PetalLength', 'PetalWidth' can improve some models.