Mathew K AnalyticsLesson 25 · Python for Retail E-commerce Analytics
Master Interpreting Retail Performance Patterns Using Python for E-commerce Analytics
In this lesson, we will solve real-world retail analytics problems by interpreting sales patterns and customer behavior in transactional data. Understanding…
- CoursePython for Retail E-commerce Analytics
- Lesson25 of 43
- Video24 min
- FormatJupyter notebook · 19 code cells
What you'll learn
- Retail Data: Concepts and Common Pitfalls
- Beginner Example 1: Calculating Total Sales Revenue
- Beginner Example 2: Counting Number of Orders
- Beginner Example 3: Identifying Top-Selling Products
- Intermediate Example 1: Analyzing Average Order Value (AOV)
- Intermediate Example 2: Sales by Country
- Intermediate Example 3: Monthly Sales Patterns
- Advanced Example 1: Identifying High-Value Customers
Data
No separate download needed — the notebook creates or downloads everything it uses.
📓 Full notebook
Download .ipynbInterpreting Retail Performance Patterns#
- In this lesson, we will solve real-world retail analytics problems by interpreting sales patterns and customer behavior in transactional data.
- Understanding sales and performance patterns helps retail businesses improve inventory management, marketing strategies, and profitability.
- We will use real e-commerce datasets to uncover actionable insights such as top-performing products, customer segments, and time-based sales trends.
- By the end, you will be able to turn raw transactional data into clear business recommendations.
import pandas as pd
import numpy as np
import warnings
warnings.filterwarnings('ignore')
Retail Data: Concepts and Common Pitfalls#
- Retail datasets often describe transactions, orders, products, and customers.
- Each row of a transactions dataset typically corresponds to a product sold in an order.
- Key sales metrics include quantity (how many sold), price (unit price), and revenue (quantity x price).
- Beginners often confuse product-level and order-level metrics, mix up gross and net revenue, or forget time granularity.
- Careful grouping and aggregation is essential for interpreting performance patterns correctly.
url = 'https://archive.ics.uci.edu/ml/machine-learning-databases/00502/online_retail_II.xlsx'
df = pd.read_excel(url, sheet_name='Year 2010-2011')
df['InvoiceDate'] = pd.to_datetime(df['InvoiceDate'])
print(df.shape)
print(df.head(3))
Beginner Example 1: Calculating Total Sales Revenue#
- The first step to interpreting retail patterns is to calculate total sales revenue.
- Revenue for each row is given by Quantity x Price.
- Total sales revenue is the sum of this value across all transactions.
df['Revenue'] = df['Quantity'] * df['Price']
total_revenue = df['Revenue'].sum()
print(f'Total sales revenue: \u00a3{total_revenue:,.2f}')
Beginner Example 2: Counting Number of Orders#
- Retailers care about how many unique orders they receive in a period.
- Each unique Invoice number typically represents one order.
- Counting unique invoice numbers gives us the total number of orders.
num_orders = df['Invoice'].nunique()
print(f'Number of unique orders: {num_orders}')
Beginner Example 3: Identifying Top-Selling Products#
- Top-selling products are those with the highest quantity sold across all orders.
- Summing quantity by product description reveals which products are most popular.
top_products = df.groupby('Description')['Quantity'].sum().sort_values(ascending=False).head(5)
print('Top 5 products by total quantity sold:')
print(top_products)
Intermediate Example 1: Analyzing Average Order Value (AOV)#
- A key metric in e-commerce is average order value: total sales revenue divided by the number of orders.
- AOV helps businesses understand customer spending behavior.
aov = total_revenue / num_orders
print(f'Average order value (AOV): \u00a3{aov:,.2f}')
Intermediate Example 2: Sales by Country#
- Retailers often sell to multiple countries, and country-wise analysis highlights key markets.
- Summing revenue per country shows which regions contribute most to sales.
sales_by_country = df.groupby('Country')['Revenue'].sum().sort_values(ascending=False).head(5)
print('Top 5 countries by sales revenue:')
print(sales_by_country)
Intermediate Example 3: Monthly Sales Patterns#
- Businesses often want to know how sales change over time.
- Summing revenue by month can highlight seasonal or trending patterns.
df['Month'] = df['InvoiceDate'].dt.to_period('M')
monthly_sales = df.groupby('Month')['Revenue'].sum()
print('Sales by month:')
print(monthly_sales.round(2))
Advanced Example 1: Identifying High-Value Customers#
- Not all customers contribute equally to sales revenue.
- Finding customers who spend the most can help with VIP marketing and retention.
top_customers = df.groupby('Customer ID')['Revenue'].sum().sort_values(ascending=False).head(5)
print('Top 5 customers by revenue:')
print(top_customers)
Advanced Example 2: Product Category Analysis (Synthetic Catalog)#
- Sometimes, you will need to join transactions to product catalog data for richer analysis.
- Let us create a synthetic product catalog and analyze which category drives most revenue.
# Create synthetic catalog
np.random.seed(42)
categories = ['Electronics','Clothing','Home','Sports','Beauty']
stock_codes = df['StockCode'].dropna().unique()
random_categories = np.random.choice(categories, size=len(stock_codes))
catalog = pd.DataFrame({'StockCode': stock_codes, 'Category': random_categories})
# Join to main data
df_cat = pd.merge(df, catalog, on='StockCode', how='left')
# Aggregate revenue per category
category_revenue = df_cat.groupby('Category')['Revenue'].sum().sort_values(ascending=False)
print('Revenue by product category:')
print(category_revenue)
Advanced Example 3: Time-Based Sales Growth Rate#
- Measuring percent change in monthly revenue shows sales growth or decline over time.
- Growth rate helps managers react quickly to performance trends.
monthly_growth = monthly_sales.pct_change().dropna() * 100
print('Monthly sales growth rate (percentage):')
print(monthly_growth.round(2))
Error Handling: Dealing With Missing Customer IDs#
- Real retail data often have missing or invalid customer IDs.
- Ignoring missing customer IDs in customer-level analysis can mislead your conclusions.
missing_customers = df['Customer ID'].isnull().sum()
total_transactions = len(df)
missing_pct = 100 * missing_customers / total_transactions
print(f'Missing customer IDs: {missing_customers} ({missing_pct:.1f}% of transactions)')
Error Handling: Correcting Incorrect Aggregations#
- Summing price instead of revenue is a common mistake.
- Always multiply quantity by price before summing for transaction-level revenue.
# Incorrect aggregation: sum of price column
wrong_total = df['Price'].sum()
print(f'Incorrect total by summing Price only: \u00a3{wrong_total:,.2f}')
# Correct total revenue is much higher
print(f'Correct total revenue: \u00a3{total_revenue:,.2f}')
Error Example: Incorrect Product or Category Grouping#
- Failing to use the correct field to group by can mix up products or categories.
- Always verify that grouping columns are uniquely identifying the product or category.
# Example: Group by 'StockCode' vs. 'Description'
group_by_code = df.groupby('StockCode')['Revenue'].sum().sort_values(ascending=False).head(3)
group_by_desc = df.groupby('Description')['Revenue'].sum().sort_values(ascending=False).head(3)
print('Grouped by StockCode:')
print(group_by_code)
print('\nGrouped by Description:')
print(group_by_desc)
Best Practice: Customer Segmentation#
- Segmenting customers by spending can help tailor marketing or loyalty efforts.
- Use quantiles to group customers into tiers based on total revenue.
customer_revenue = df.groupby('Customer ID')['Revenue'].sum()
segments = pd.qcut(customer_revenue, q=4, labels=['Bronze','Silver','Gold','Platinum'])
segment_counts = segments.value_counts()
print('Customer segmentation by revenue tier:')
print(segment_counts)
Best Practice: Product Performance Analysis#
- Always compare both total quantity sold and total revenue by product.
- Some products are bestsellers by volume but not by revenue (and vice versa).
product_performance = df.groupby('Description').agg({'Quantity':'sum','Revenue':'sum'})
print('Sample product performance rows:')
print(product_performance.sort_values('Quantity', ascending=False).head(3))
print(product_performance.sort_values('Revenue', ascending=False).head(3))
Best Practice: Detecting Seasonal Trends#
- Identifying months with sales peaks or dips guides promotional campaigns.
- Overlaying category or product sales by month can reveal product-specific seasonality.
# Example: Category revenue by month
df_cat['Month'] = df_cat['InvoiceDate'].dt.to_period('M')
cat_month = df_cat.groupby(['Category','Month'])['Revenue'].sum().unstack(0).fillna(0)
print('Revenue by category and month (sample):')
print(cat_month.head())
End-to-End Example: From Transactions to Business Insight#
- Let us answer a critical retail question: Which product category should the business prioritize for upcoming promotions?
- We will combine revenue totals and seasonality to make a clear recommendation.
# Step 1: Find top revenue-driving category
top_category = category_revenue.idxmax()
print(f'Top category by total revenue: {top_category}')
# Step 2: Find peak month for this category
cat_peak_month = cat_month[top_category].idxmax()
print(f'Peak sales month for {top_category}: {cat_peak_month}')
Found this useful?
All lessons, notebooks and datasets here are free. If they helped you, a coffee keeps new lessons coming.
Related lessons
23 minLesson 25
Interpreting Patterns and Market Trends in Python Market Research Analytics
27 minLesson 47
Analyzing Market and Consumer Trends Over Time Using Python | Market Research Analytics
25 minLesson 34
Chi-Square Tests for Analyzing Survey Relationships Using Python | Market Research Analytics
22 minLesson 33