Skip to main content

Feature Engineering Pipeline

Systematic feature engineering including selection, transformation, encoding, scaling, and dimensionality reduction with feature importance analysis.

Complexity: Moderate | Duration: 15-30m | Category: Devops

Tags: ml feature-engineering preprocessing

Workflow Diagram​

Steps​

Step 1: Feature discovery​

Agent: data

scientist - Identify candidate features from raw data

Step 2: Feature encoding​

Agent: data

scientist - One-hot, label, target encoding for categoricals

Step 3: Feature scaling​

Agent: ml

engineer - StandardScaler, MinMaxScaler, or RobustScaler

Step 4: Feature transformation​

Agent: data

scientist - Log, sqrt, polynomial, interaction features

Step 5: Feature selection​

Agent: ml

engineer - Mutual info, LASSO, tree-based importance

Step 6: Dimensionality reduction​

Agent: ml

engineer - PCA, t-SNE, UMAP if high-dimensional

Step 7: Feature validation​

Agent: data

scientist - Check for leakage, multicollinearity, variance

Step 8: Feature documentation​

Agent: data

scientist - Document feature definitions and transformations

Usage​

To execute this workflow:

/workflow devops/feature-engineering-pipeline.workflow

See other workflows in this category for related automation patterns.