Loading the source tables…

Walmart Sales Intelligence

— Showing all stores

Weekly revenue

filtered

Revenue by format

filtered

Week of year, all years stacked

filtered

Revenue by archetype

filtered

Top 15 departments

filtered

Top 15 stores

filtered

Holiday lift by holiday

full dataset

Biggest weeks, and whether flagged

full dataset

Revenue concentration

full dataset

Findings

full dataset

Every store, shared scale

filtered · click to select

Floor space against revenue per sq ft

filtered

Revenue against volatility

filtered

Archetype map, principal components

filtered

Choosing k

full dataset

Archetype profiles

full dataset

Store-week spread by format

full dataset

Store scorecard

filtered · click a row

Store	Fmt	Archetype	Size	Revenue	Avg week	$/sqft	CV	Hol lift	Growth	Unemp	Gap

Productivity headroom, largest gaps

filtered

Department Pareto

full dataset

Department share

full dataset

Holiday lift by department

full dataset

Growth, 2011 on 2010

full dataset

Store against department, share of each store

filtered

Markdown depth against sales

full dataset

Net returns by department

full dataset

All departments

full dataset

Dept	Revenue	Share	Cumulative	Avg week	CV	Stores	Hol lift	Growth	Returns

Missing values by column

raw merge

Missing markdowns over time

raw merge

Distribution of a column

raw merge

Skew and tail weight

raw merge

Outlier share by column

raw merge

Spread of every numeric column

raw merge

Sales across deciles of each driver

raw merge

Record counts by category

raw merge

Correlation

full dataset

Column profile

raw merge

Column	Role	Type	Missing	Unique	Mean	Median	Std	Min	Max	Skew	Kurtosis	Outliers	Negatives

Data quality checks

raw merge

Check	Value	Note

Model fit, three specifications

full dataset

Predicted against actual

full dataset

Residuals

full dataset

Impact of a one-sd move

full dataset

Sampling error against sample size

full dataset

Sampling distribution of the mean

full dataset

Observed against fitted densities

full dataset

Normal Q-Q

full dataset

Goodness of fit, KS statistic

full dataset

Percentile ladder

full dataset

Holiday against normal weeks

full dataset

Probability of clearing a threshold

full dataset

Mean sales by format, with 95% interval

full dataset

The eight experiments

full dataset

#	Experiment	Method	Headline result

Built from the Walmart Store Sales Forecasting dataset: 421,570 store-department-week records across 45 stores and 81 departments, 5 Feb 2010 to 26 Oct 2012. Every figure is read from a table in docs/data/; nothing is recomputed in the browser. Source and the analysis pipeline: github.com/jeet-1236/walmart-sales-intelligence.

Walmart Sales Intelligence

Weekly revenue

Revenue by format

Week of year, all years stacked

Revenue by archetype

Top 15 departments

Top 15 stores

Holiday lift by holiday

Biggest weeks, and whether flagged

Revenue concentration

Findings

Decomposition

Seasonal index

Revenue by month

14-week forecast

Backtest, 13 held-out weeks

Forecast accuracy

Revenue by year

Sales against each driver

Every store, shared scale

Floor space against revenue per sq ft

Revenue against volatility

Archetype map, principal components

Choosing k

Archetype profiles

Store-week spread by format

Store scorecard

Productivity headroom, largest gaps

Department Pareto

Department share

Holiday lift by department

Growth, 2011 on 2010

Store against department, share of each store

Markdown depth against sales

Net returns by department

All departments

Missing values by column

Missing markdowns over time

Distribution of a column

Skew and tail weight

Outlier share by column

Spread of every numeric column

Sales across deciles of each driver

Record counts by category

Correlation

Column profile

Data quality checks

Model fit, three specifications

Predicted against actual

Residuals

Impact of a one-sd move

Sampling error against sample size

Sampling distribution of the mean

Observed against fitted densities

Normal Q-Q

Goodness of fit, KS statistic

Percentile ladder

Holiday against normal weeks

Probability of clearing a threshold

Mean sales by format, with 95% interval

The eight experiments