Exploratory Data Analysis Report: Ecommerce Sales Dataset

Report Generated: /workspace/webserver/lib/smartvox/skill_hub/eda_report_ecommerce_sales_20251203_180016.html

Dataset Overview

• Size: 13,979 rows × 4 columns
• Memory: 1.94 MB
• Completeness: 100% (no missing values)
• Data Quality: Excellent (0 duplicates, 0 missing values)

Columns:

• Country Name (categorical): 262 unique countries/regions
• Country Code (categorical): 262 unique codes
• Year (numeric): 1960-2023 (64 years coverage)
• Value (numeric): Economic/sales values

Key Patterns

1. Extreme Growth Over Time

• 1960 Starting Point: Average value of $72B per country/region
• 2023 Endpoint: Average value of $3.7T per country/region
• Overall Growth: 5,083% increase over 64 years
• Pattern: Exponential growth trajectory, reflecting global economic expansion

2. Top Economic Regions (by Total Value)

1. World: $2.15 quadrillion total
2. High income countries: $1.57 quadrillion
3. OECD members: $1.51 quadrillion
4. Post-demographic dividend: $1.43 quadrillion
5. Europe & Central Asia: $662 trillion

3. Top Regions by Average Value

1. World: $33.6 trillion
2. High income: $24.6 trillion
3. OECD members: $23.5 trillion
4. Post-demographic dividend: $22.4 trillion
5. Europe & Central Asia: $10.4 trillion

4. Data Coverage

• Balanced: 262 countries/regions tracked consistently
• Top 10 entries: 64 observations each (complete coverage across all years)
• Complete timeline: Full 1960-2023 coverage for most entities

Statistical Insights

Value Distribution

• Mean: $1.21 trillion (heavily influenced by large economies)
• Median: $16.7 billion (more typical country value)
• Large Gap: Mean >> Median indicates extreme right skew

Distribution Characteristics

• Skewness: 8.35 (highly right-skewed)
  • Dominated by large economies (World, High income, OECD)
  • Most countries have much smaller values
• Kurtosis: 90.09 (extreme outliers present)
  • Heavy tails with very large economies far from the center

Outliers

• Count: 2,311 observations (16.53% of dataset)
• Pattern: Expected outliers are aggregated regions and major economies
• IQR Bounds: -$303B to $511B
  • Values above upper bound include developed nations and regional aggregates
  • Outliers are data-rich, not errors

Data Quality Assessment

Strengths

• Perfect completeness (0% missing data)
• No duplicate records
• Consistent temporal coverage (1960-2023)
• Balanced representation across countries/regions

Characteristics

• Dataset includes both individual countries AND regional aggregates (e.g., “World”, “High income”, “OECD members”)
• This explains the extreme outliers and high skewness
• Regional aggregates inflate mean statistics

Recommendations

For Analysis

1. Segment the data:

   • Separate individual countries from regional aggregates
   • Analyze small/developing vs large/developed economies separately

2. Consider log transformation:

   • Extreme skewness (8.35) suggests log-scale analysis for Value
   • Would normalize distribution for modeling

3. Time series analysis:

   • Strong temporal component (64 years)
   • Analyze growth rates rather than absolute values
   • Consider inflation adjustment if comparing across decades

4. Outlier treatment:

   • Keep outliers (they’re legitimate data points)
   • But flag regional aggregates vs individual countries for separate analysis

For Visualization

1. Use logarithmic scales for Value comparisons
2. Create separate views for countries vs regional aggregates
3. Show growth rates (YoY %) rather than absolute values
4. Consider interactive filtering by income level or region

For Modeling

1. Feature engineering: Calculate growth rates, decade indicators
2. Stratify by country type (individual vs aggregate)
3. Account for temporal dependencies (time series methods)
4. Consider GDP per capita or relative measures instead of absolute values

Critical Finding

This dataset appears to represent global economic data (likely GDP or similar economic indicator) rather than traditional “ecommerce sales.” The presence of aggregated regions like “World”, “High income”, “OECD members” and the 64-year historical span (1960-2023) are characteristic of World Bank or IMF economic datasets, not retail ecommerce transactions.