Target Variables
Figure 1: CCI Distribution
CCI is tightly centered around 100 because the OECD series is amplitude-adjusted and smoothed.
Figure 2: Retail Distribution
Retail index varies more strongly by country and period, making it easier to predict than CCI.
Figure 3: Correlation Matrix
Weather variables are strongly correlated with each other, which justifies PCA before linear modeling.
Categorical and Time-Series Views
Figure 4: Categorical vs Target
Seasonal differences are visible, but CCI seasonal tests remain statistically non-significant.
Figure 5: Weather Time Series
Sunshine has a strong seasonal cycle, while the consumer outcomes respond less directly.
Notebook EDA Figures
