Time-Series Revenue Analysis with Confidence Bands
Daily revenue with 7-day moving average and 95% confidence intervals.
The shaded region represents statistical uncertainty in the trend estimation.
Outlier Detection: Multiple Statistical Methods
Comparing three robust outlier detection methods: Z-score (|z| > 2), IQR (1.5×IQR rule),
and Modified Z-score using MAD (|modified z| > 3.5). Outliers indicate days with exceptional deviation from expected behavior.
Trend Decomposition
Polynomial trend (degree 2) fitted to revenue data. Deviations from trend reveal
short-term fluctuations and seasonal patterns.
Coefficient of Variation Over Time
Rolling 7-day CV (%) measures relative variability. Higher CV indicates greater
uncertainty and less predictable revenue patterns.
Segment Comparison: Category Performance
Mean sales per order with 95% confidence intervals. Non-overlapping intervals suggest
statistically significant differences between segments.
Segment Comparison: Channel Performance
Distribution statistics across sales channels. Standard error and confidence intervals
quantify estimation uncertainty.
Segment Comparison: Customer Type Performance
Statistical comparison of customer segments. Coefficient of variation reveals
within-segment variability.
Promotion Effectiveness: Statistical Hypothesis Test