Historical target
The saved model predicts a dataset field called “not fully paid.” That label does not establish when a payment was missed or whether a loan was formally in default.
INTERACTIVE MODEL EXPLORERUnderstand a historical
Understand a historical
repayment signal.
Explore how the saved machine learning model responds to a complete fictional loan scenario. Every input is visible. Results are estimates from historical data, with limits shown alongside them.
19 model features0 scenarios stored1 historical label
01 / WORKSPACE
Build a scenario
Start with the fictional example and change any input. Do not enter a real person's details.
02 / MODEL NOTES
Know what the output means
The model is useful as a portfolio demonstration when its boundaries remain visible.
Complete input mapping
The original app silently set omitted model fields to zero. This version collects every numerical field and maps one purpose category to the seven encoded columns.
Evidence still needed
The source dataset, reproducible evaluation, calibration, fairness review and current population checks must be supplied before any operational assessment.