HongZe Project

Project step: 📌 Rebar Demand Forecasting Project Plan

1️⃣ Data Cleaning & Manipulation ✅ (Completed) §

• Merged multiple datasets (df_rebar, df_cement, df_real_estate, df_infra).

• Handled missing values (NaN and NaT).

• Aligned timestamps and ensured proper date formatting.

• Aggregated weekly rebar prices and ensured consistency.

2️⃣ Exploratory Data Analysis (EDA) & Regression Analysis 🔄 (In Progress) §

• Generated descriptive statistics for key variables.

• Created correlation heatmaps to identify important relationships.

• Conducted multiple linear regression to analyze variable influence on rebar_demand.

• Evaluated regression model performance using R², MSE, and visualization.

3️⃣ Feature Engineering ❗ (Next Step - Recommended Before Time Series Analysis) §

• Lag Features: Add past rebar_demand values as new predictors.

• Rolling Window Features: Compute moving averages for smoother trends.

• Seasonality Encoding: Apply Sin-Cos encoding for month & week_of_month to capture cyclic behavior.

• Interaction Terms (Optional): Explore relationships between economic indicators and demand.

4️⃣ Time Series Analysis 🔜 (Upcoming) §

• Check stationarity using the ADF test.

• Identify seasonal patterns and trends.

• Perform time series decomposition (Trend, Seasonal, Residuals).

• Visualize historical demand trends and detect anomalies.

5️⃣ Modeling & Forecasting 🔜 §

• Test different models for predicting rebar_demand:

  • Facebook Prophet (trend + seasonality modeling).

  • ARIMA/SARIMA (statistical forecasting).

  • XGBoost / CatBoost (machine learning-based predictions).

  • LSTM / GRU (deep learning for long-term time series forecasting).

• Tune hyperparameters for optimal model performance.

6️⃣ Model Evaluation & Optimization 🔜 §

• Compare models using:

  • Root Mean Squared Error (RMSE)

  • Mean Absolute Percentage Error (MAPE)

  • Cross-validation scores

• Select the best model based on accuracy & generalization.

• Fine-tune model parameters for better predictions.

Final Deliverables 📄 §

• Cleaned dataset with engineered features.

• Exploratory Data Analysis (EDA) report.

• Regression analysis results (variable importance).

• Time series analysis & forecasting models.

• Model evaluation & comparison report.

---

🚀 Next Steps: ✅ Complete EDA & Regression Analysis 🔜 Implement Feature Engineering before moving to Time Series Analysis 📊 Proceed to Modeling & Forecasting

---

📌 Let me know if you’d like to add any specific steps or focus on certain models! 🎯