Case Study | December 2025
π· Wine & Weather
Do Rising Temperatures Drive Summer Wine Sales?
A comprehensive statistical investigation testing the global warming hypothesis on 178 months of Norwegian alcohol sales data using Power BI MCP and Statistics MCP.
The Hypothesis
Claim: Rising temperatures due to global warming are pushing increased sales of White wine (Hvitvin), RosΓ© wine (RosΓ©vin), and Sparkling wine (Musserende Vin) in Norway.
This investigation combines live Power BI semantic model data with rigorous statistical testing to evaluate this hypothesis through multiple analytical approaches.
Analysis Workflow
Power BI Connection
Connected to SalesVP Power BI semantic model. Extracted monthly wine sales for White, RosΓ©, and Sparkling categories.
Data Integration
Merged with Norwegian meteorological data including monthly temperature averages and standardized anomaly scores (Ο from historical mean).
Correlation Analysis
Pearson and Spearman correlations between temperature anomalies and wine sales. Summer seasonal analysis (May-August).
Hypothesis Testing
Welch's t-tests comparing warm vs cold anomaly months. Cohen's d effect size calculations.
Regression Modeling
Multiple regression controlling for time trend, seasonality, and temperature anomaly. RΒ² and coefficient significance.
Quarto Documentation
Generated publication-ready statistical report with dynamic Python visualizations and inline computed statistics.
Statistical Analysis Results
Comprehensive statistical testing using MCP Statistics tools:
| Analysis | Method | Result | Conclusion |
|---|---|---|---|
| Overall Correlation | Pearson r |
r = 0.18, p = 0.018 | β Significant (weak) |
| Rank Correlation | Spearman Ο |
Ο = 0.16, p = 0.032 | β Significant (weak) |
| Summer Correlation | Pearson r |
r = 0.52, p = 0.047 | β Significant (moderate) |
| Summer Temp Trend | Linear Regression |
Ξ² = +0.034 SD/year, RΒ² = 28.5% | β Warming trend |
| Warm vs Cold Months | Welch's t-test |
t = 1.42, p = 0.158 | β Not significant |
| Effect Size | Cohen's d |
d = 0.08 | β οΈ Negligible effect |
Evidence Summary
β Supporting Evidence
- Moderate summer correlation (r = 0.52, p = 0.047) between annual temperature anomalies and wine sales
- Clear warming trend in summer temperatures (+0.034 SD/year, p = 0.041)
- Weak but significant overall correlation (r = 0.18, p = 0.018)
- Both Pearson and Spearman confirm positive relationship
β οΈ Limiting Factors
- No significant warm vs cold difference (t = 1.42, p = 0.158)
- Very small effect size (Cohen's d = 0.08, negligible)
- COVID-19 (2020-2021) creates major outliers
- Time/market trend dominates temperature effect
- Correlation β causation
The Tech Stack
π€ GitHub Copilot
Claude orchestrating the analysis workflow
π Power BI MCP Server
Connect to Vinmonopolet semantic model via DAX
π MCP Statistics
Correlation, t-tests, regression, effect sizes
π Python + Matplotlib
Dynamic visualizations with publication styling
π Quarto
Reproducible document with inline statistics
π‘οΈ Met.no Data
Norwegian meteorological temperature records
Key Takeaways
Rigorous Hypothesis Testing
Multiple statistical methods applied to evaluate a real-world business hypothesis with proper effect size consideration.
Power BI + Statistics Integration
Seamless workflow from live semantic model data to comprehensive statistical analysis in a single session.
Honest Conclusions
Statistical significance without practical significance. Effect sizes matter as much as p-values.
The Result
A complete statistical investigation with dynamic visualizations and reproducible analysis:
- β 8 publication-quality visualizations
- β 6 statistical tests with effect sizes
- β Multiple regression with 3 predictors
- β COVID-19 confounding analysis
- β Dynamically computed inline statistics
Explore More
See how MCP tools enable sophisticated analytics from your existing data sources.