PranavSharma/turnover-forecasting-model

📊 AI-Powered Turnover Forecasting for SAP SE

🚀 Project Overview

This project delivers AI-driven revenue forecasting for SAP SE using a univariate SARIMA model. It shows how accurate forecasts can be built from limited data (just historical turnover).

---

🏢 Why SAP SE?

• SAP SE is a global leader in enterprise software
• Revenue forecasts support strategic planning & growth
• Perfect case for AI-powered financial forecasting

---

🧠 Model Details

• Model type: SARIMA (Seasonal ARIMA)
• Trained on: SAP SE revenue from Top 12 German Companies Dataset (Kaggle)
• SARIMA Order: (3, 1, 5)
• Seasonal Order: (0, 1, 0, 12)
• Evaluation Metric: MAE (Mean Absolute Error)
• Validation: Walk-forward validation with test set (last 10%)

---

⚙️ How to Use

import pickle

with open("sarima_sap_model.pkl", "rb") as f:
    model = pickle.load(f)

forecast = model.forecast(steps=4)
print(forecast)

📌 Intended Use & Limitations

👍 Forecast SAP SE revenue for next 1–6 quarters
📈 Great for univariate, seasonal time series
🚫 Not suitable for multivariate or non-seasonal data
⚠️ Requires careful preprocessing (e.g., stationarity)

👨‍💻 Author: Pranav Sharma