Validated row counts, dates, duplicates and complete store histories across SQL and Python
VERIFIED CASE STUDY · RETAIL PERFORMANCE
Walmart Sales Trend · 2010–2012
An end-to-end retail analysis of 45 Walmart stores that corrects a misleading holiday flag and shows when sales peak, which stores drive performance and what should inform the forecast.
SQLPythonExcelTableau
01 · CONTEXT
From a broad concern to testable questions.
Business problem
How should regional leadership plan inventory, staffing and promotions when the raw holiday calendar misses the biggest sales week of the year?
Questions explored
- Which holidays actually lift sales, and by how much?
- Which stores generate revenue and which are gaining or losing momentum?
- Does the holiday peak vary by store?
- Do temperature, fuel price, CPI or unemployment explain weekly sales?
Dataset
Walmart weekly sales dataset covering 45 stores, 143 weeks and 6,435 complete store-week records from 5 February 2010 to 26 October 2012.
02 · ANALYSIS
A transparent path from preparation to interpretation.
Corrected the holiday calendar after finding that the source missed the pre-Christmas peak and flagged the week after Christmas
Built ten SQL analyses with CTEs, window functions, Pareto logic and Pearson correlation
Cross-checked every result in pandas and formula-driven Excel before designing Tableau and web dashboards
03 · DASHBOARD
TABLEAU + WEB REPORT GALLERY1 decision view
Select any view to inspect the dashboard at full size.
04 · FINDINGS
What the analysis surfaced
01The pre-Christmas week averaged $79.0M—70.3% above a regular week—but was not flagged as a holiday; the week marked Christmas was 6.7% below normal.
02Only pre-Christmas (+70.3%) and Thanksgiving (+42.8%) produced material lift; Super Bowl and Labor Day behaved close to regular weeks.
03Thirty-seven of 45 stores grew in 2012, but Store 14—Walmart's third-largest store in this data—lost $7.3M and accounted for about half the decline across eight falling stores.
04Temperature, fuel price, CPI and unemployment had correlations between −0.11 and +0.01 with weekly sales, far weaker than store history and calendar effects.
Recommended next moves
01Fix the reporting and forecasting calendar so Thanksgiving through 24 December is treated as the peak period.
02Concentrate staffing, stock and promotion budgets on the two weeks that produce genuine holiday lift, then plan lighter January schedules and clearance activity.
03Review Store 14 and Store 36 operationally, while studying Stores 38, 44 and 39 for practices that can transfer.
04Allocate holiday inventory using each store's own 1.0x–2.05x peak multiplier and forecast from store history and calendar effects rather than weak macro variables.
LIMITATIONS
What the analysis does not establish.
- The history covers 2.75 years and only two complete holiday seasons, so event lift estimates rely on two to three observations.
- The dataset has no product, department, margin, traffic or store-size fields, limiting diagnosis of why individual stores changed.
- Correlation does not establish causation, and the weak macro relationships apply only to this period and store mix.
- The fixed pre-Christmas date rule should be validated against future years before production forecasting.
NEXT CASE STUDY
Continue exploring