Pipeline Stages
- Data Loader
- Imputation
- Encoding
- Target Selector
- Target / Cluster Distribution
- Train/Test Split
- Logistic Regression
- Metrics
- Confusion Matrix
- ROC / PR Curve
Data Loader. Loads the Telco Customer Churn dataset (7043 samples, 20 features). Source: IBM sample data, telco-customer-churn-on-icp4d. License: Apache 2.0.
Imputation. Median fill. Eleven total-charge cells are blank, and the fill is learned on the train rows.
Encoding. One-hot encoding. customerID and Churn are excluded.
Target Selector. Target column is Churn.
Target / Cluster Distribution. Before the split, 26.5 percent of 7,043 customers left.
Train/Test Split. Holds out 1,409 rows, 20 percent, seed 42, stratified.
Logistic Regression. C 1.0, lbfgs, class weight off. The node standardises the train rows because no scaler is on the canvas.
Metrics. Hold-out accuracy 0.8062. For customers who left, precision 0.6573, recall 0.5642, F1 0.6072.
Confusion Matrix. Rows are the actual class. The Yes row is the customers who left.
ROC / PR Curve. ROC-AUC 0.8416. Average precision 0.6308.
