Customer Churn Prediction — Sandbox

Train a churn-risk model from historical customers, upload current customers, score churn probability, identify the strongest risk drivers, and export a retention queue for CRM or sales teams. The browser demo mirrors the converted Flask workflow and keeps uploaded data local to the page.

Best workflow: load training data → select churn target → train → load scoring data → score customers → review high-risk drivers → export retention list.

Column guide: the churn target should contain values such as Yes/No, 1/0, Churned/Retained, or True/False. ID-like columns are automatically ignored.

Upload historical customer CSVMust include customer features and a churn/attrition label.

Churn target column

Pick the column that tells the model who churned and who stayed. Example: Churned.

Columns to exclude

Tick customer IDs, leakage fields, or fields you do not want used in training. The Output Dashboard updates immediately to show the selected exclusions.

Upload current customer CSVNo churn column needed. The trained model appends probability, risk level, drivers, and action.

High-risk threshold

More leads50%Stricter

Churn assistant ready.

After scoring, ask questions like “Which customers need urgent retention action?”, “What are the strongest churn drivers?”, or “What is the ROI opportunity?”

SummaryDriversActionsROI

Rows Loaded

Columns Loaded

Excluded

Model Accuracy

—

High Risk

—

Dataset Preview

Switch between training, scoring, excluded-feature view, or final predictions. Preview top or bottom rows like a dataframe.

Status
Load sample data or upload a CSV to preview records here.

Risk Distribution

Run scoring to generate the High / Medium / Low risk distribution.

Top Risk Drivers

Train the model to see the most influential churn drivers.

ROC-AUC

—

Precision View

—

Revenue at Risk

—

Retention Opportunity

—

Interpretation: the sandbox estimates churn risk, highlights drivers, and prioritises customers for retention. In production, the Flask app can connect to CRM, BI dashboards, or a scheduled batch-scoring workflow.

Customer Churn Prediction — Sandbox

Output Dashboard

Dataset Preview

Risk Distribution

Top Risk Drivers

Model Performance & Retention Value