File size: 1,416 Bytes
5b21e2e fbcf676 e2eb1bd 5b21e2e fbcf676 5b21e2e fbcf676 5b21e2e e2eb1bd 5b21e2e e2eb1bd 5b21e2e e2eb1bd | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 | ---
title: EmployeeAttritionClassification
emoji: 🏢
colorFrom: indigo
colorTo: blue
sdk: docker
app_port: 8501
tags:
- streamlit
pinned: false
short_description: Employee Attrition Predictor using XGBoost.
license: mit
---
## Overview
This app predicts the **probability of employee attrition (Attrition = 1)** using a trained **XGBoost classifier**.
The model was evaluated using **ROC-AUC** and achieved strong validation performance.
## Files in this repo
Place these files in the **repo root (same folder as `app.py`)**:
- `xgb_model.pkl` — trained XGBoost model
- `feature_names.pkl` — list of training feature columns (order matters)
- `threshold.pkl` — decision threshold (e.g., 0.35)
## How to use the app
### 1) Single prediction (form)
- Enter numeric/ordinal feature values
- Select one option for categorical groups (one-hot)
- The app outputs:
- `Attrition probability`
- `Attrition prediction` using your saved threshold
### 2) Batch prediction (CSV upload)
Upload a CSV that is already in the **same feature format** as your training data
(after preprocessing and one-hot encoding).
The app will align columns automatically using `feature_names.pkl`.
## Notes
- Kaggle submissions typically require **probabilities**, not class labels.
- If you want to change the decision threshold, update `threshold.pkl`.
## Local run
```bash
pip install -r requirements.txt
streamlit run app.py |