Real Client Project · UAE
Fuso Demand Dashboard
Multi-model demand forecasting for a UAE commercial vehicle distributor — replacing manual Excel S&OP with live, decision-ready analytics.
18%MAPE Reduction
3Forecast Models Compared
24moHistorical Data Window
6moForward Horizon
12SKU Categories
Demand Forecast vs Actual
6-month rolling forecast with confidence intervals — ensemble model output
Model Performance Comparison
Automatic model selection picks the best performer per SKU category
XGBoost Ensemble
11.2%
✓ Best Overall
Handles non-linear seasonality and UAE market shocks
Holt-Winters ETS
13.7%
Good for Seasonal
Strong on Ramadan & National Day cyclical patterns
ARIMA (Auto)
15.4%
Good for Stable
Reliable baseline for stable commercial vehicle lines
Naïve Baseline
29.3%
Previous Method
Manual Excel estimates before dashboard deployment
SKU-Level Forecast Summary
Top vehicle categories — Q3 forecast vs prior quarter actual
Tech Stack
Tools and technologies used to build and deploy this dashboard
Python (pandas, statsmodels, scikit-learn)
Power BI
ARIMA / Auto-ARIMA
Holt-Winters ETS
XGBoost
SQL Server
Excel (XLOOKUP, Power Query)
UAE Automotive Market Data