📈 Procurement Spend Forecaster

Upload historical PO data → ML predicts future procurement spend with multiple models.

Built by Ekhsan Fitri — demonstrating time-series ML forecasting for business applications.

---

✨ Features

• 📂 Upload Excel/CSV — auto-detect date & amount columns
• 🧠 3 ML models — Linear Regression, Random Forest, Gradient Boosting
• 🔮 Ensemble forecast — combined prediction for robustness
• 📊 Confidence bands — ±15% prediction interval
• 📈 Trend detection — automatically identifies spend trends
• ⚡ Weekly/Monthly/Quarterly — flexible time granularity
• 📦 Sample data — 24 months with trend + seasonality + noise

🚀 Quick Start

git clone https://github.com/EkhsanFitri94/procurement-forecaster.git
cd procurement-forecaster
pip install -r requirements.txt
streamlit run app.py

📸 Demo

streamlit run app.py
# → Upload procurement Excel/CSV (or click "Load Sample Data")
# → Select Date + Amount columns
# → Instant ML forecast with 3-model ensemble

Dashboard shows:

• 📈 Historical spend + forecast line with ±15% confidence band
• 🧠 3 models compared: Linear Regression, Random Forest, Gradient Boosting
• 📊 Model accuracy (MAE) bar chart
• 🔮 Forecast values table for future periods
• 📦 24 months of sample data with trend + seasonality pre-loaded

💡 Works with any time-series procurement data — monthly, weekly, or quarterly.

🎯 Skills Demonstrated

• Time-series feature engineering
• Multi-model training & comparison
• Ensemble forecasting
• Interactive data visualization (Plotly)
• Business ML application

---

Part of Ekhsan Fitri's AI & Procurement portfolio · More projects