Analysis Overview
Analysis overview and configuration
| Parameter | Value | _row |
|---|---|---|
| confidence_level | 0.95 | confidence_level |
| forecast_horizon | 30 | forecast_horizon |
| seasonal_period | 7 | seasonal_period |
Data preprocessing and column mapping
Executive Summary
Executive summary with key findings and actionable recommendations
| Metric | Value |
|---|---|
| Data Period | 4,826 daily observations |
| Best Model | ETS |
| Forecast Accuracy (MAPE) | 30.8% |
| Trend Direction | Increasing |
| Seasonal Strength | 0.3 (Moderate) |
| Forecast Horizon | 30 days ahead |
Key Findings:
• Trend: Sales show a increasing trend pattern
• Seasonality: Moderate seasonal strength (0.30) - seasonal effects present but mixed with trend
• Best Model: ETS outperformed alternatives with 30.8% error
• Forecast Horizon: 30-day ahead predictions with 95% confidence intervals
Recommendation: Use the ets forecast for operational planning. Apply the 95% upper confidence bound for safety stock calculations to minimize stockout risk. Re-train the model weekly with recent data to maintain forecast accuracy as patterns evolve.
Historical Trends
Historical sales with moving averages showing trend and volatility
STL Decomposition
STL decomposition showing trend, seasonal, and residual components
Weekly Seasonality
Day-of-week sales distribution showing weekly patterns
ACF/PACF Diagnostics
Autocorrelation and partial autocorrelation plots for model diagnostics
Forecast Results
Best model (ETS) forecast with 95% confidence intervals
Model Comparison
Performance comparison across all forecasting models tested
Residual Diagnostics
Residual distribution analysis with normality assessment
Q-Q Plot
Residual distribution analysis with normality assessment
Actual vs Predicted
Actual vs predicted scatter plot showing forecast accuracy on test data
Store Comparison
Demand comparison across stores showing volume leaders and laggards
Item Demand Ranking
Item demand ranking showing top sellers and slow movers
Demand Heatmap
Store x item demand matrix showing cross-sectional patterns