Click the link to use the template.
E-commerce has transformed the way we shop, becoming a crucial sales channel for businesses worldwide. What insights lie hidden in the vast data generated by e-commerce transactions? How can we leverage this user data to help e-commerce companies reduce costs, improve efficiency, and enhance customer service?
This case will demonstrate how to harness AI to analyze e-commerce user data, enabling companies to better understand user behavior and market trends.
Dataset preparation
Dataset Introduction: This dataset offers comprehensive information on toy products, enabling an in-depth analysis of consumer purchasing and market dynamics. Researchers and market analysts can use this data to gain insights into consumer preferences, assess product performance, and develop or optimize sales strategies. The dataset includes essential product details as well as critical market data such as price, inventory, and star ratings.
First, we execute the following AI block for data processing.
prompt:
Delete duplicate rows from the table, and write the results to the table
unique id: unique identifier for each product, used to distinguish different products.
product name: name of the product, describes the basic information of the product.
manufacturer: provides the name of the company that produces the product.
price: the price of the product, reflecting the market pricing strategy.
number available in stock: the number of available stock, indicating the availability of the product.
number of reviews: number of user reviews, indicating the popularity of the product.
number of answered questions: the number of questions answered, reflecting the degree of customer attention to the product.
average review rating: average review star rating, provides an indicator of consumer satisfaction with the product.
amazon category and sub category: the Amazon category and sub-category that the product belongs to, which helps market positioning analysis.
customers who bought this item also bought: users who bought this product also bought other items, revealing consumer buying habits.
description: product description, providing detailed information about the product.
product information: other related product information, covering technical parameters and so on.
product description: detailed product description, used to attract potential customers.
items customers buy after viewing this item: other items purchased by the customer after viewing this product, indicating the buying decision process.
User behavior analysis
Analytical Objectives
In this demo's hands-on instruction, we focus on Exploratory Data Analysis (EDA) to help users practice techniques such as summary statistics, visualization, and identifying patterns in data.
Analysis Tasks - Exploratory Analysis (EDA)
Let's start with some simple data analysis.
Price analysis
Analyze the distribution of product prices to grasp the pricing range within the market.
prompt:
Display the number of products in different price ranges based on price. Line Chart
AI Chart Interpretation
The chart displays the number of products available in various price ranges, showing a clear decreasing trend as the price range increases. The majority of products are priced between $0-10, with 342 products, and the number diminishes significantly in higher price ranges. Notably, there are very few products over $200, with virtually none in the $1000+ range.
Inventory analysis
Evaluate product inventory levels and identify risks of stock-outs or surpluses.
prompt:
Evaluate the inventory level of a product based on the number_available_in_stock field. Draw a bar chart showing the top 15 products with the most inventory, in descending order
AI Chart Interpretation
The bar chart displays the top 15 products with the most inventory available in stock. The "LEGO Star Wars 75097: Advent Calendar" has the highest inventory at 73 units, closely followed by "LEGO Friends 41102 Advent Calendar" with 72 units. The inventory levels for the remaining products gradually decrease, with the lowest being "Kato 23-226 Yard Buildings Kit" at 34 units. This indicates a notable concentration in inventory for the top two LEGO products compared to the rest.
Brand Analysis
To study the product performance and market acceptance of different manufacturers.
prompt:
Top 15 Manufacturers by Number of Products, Pie Chart
AI Chart Interpretation
The pie chart displays the top 15 manufacturers by the number of products available. Hornby leads the list, accounting for 19.6% of the products, followed by Q-Workshop with 11.2%. Oxford Diecast, Corgi, and Kato also have notable shares, at 8.7%, 7.4%, and 7.0%, respectively. The smallest shares, each at 3.86%, are shared by LEGO, Mattel, and Steiff.
Market Pricing Strategy Analysis
prompt:
Based on price price number of reviews average review rating , plot violin plot to show the distribution of product reviews in different price ranges and analyze the impact of price on reviews. Violin plot horizontal coordinate for the price range (0,10) (10,20) (20,50) (50,100) (100,200) (200,500) (500.1000), vertical coordinate for the average review rating (from 3.5 4.0 to 5.5)
AI Chart Interpretation
The violin plot shows the distribution of product reviews across various price ranges. Here are some observations:
Diverse Ratings Across Price Ranges: Products in the lowest price range (0-10) typically have a concentrated distribution around higher ratings (around 4.5), whereas higher price ranges (50-100 and above) show more variability and some outliers with both high and low ratings.
Moderately Priced Products: Price ranges from 10-50 tend to have well-distributed ratings, mostly centered around 4.5, indicating a consistent level of satisfaction with products in this range.
High Price Variability: The distribution becomes narrower in the highest price ranges (200-1000), with some extreme outliers, suggesting that while many products are well-reviewed, a few receive significantly lower ratings, indicating potential dissatisfaction for some expensive items.
Product Description Attractiveness Analysis
Evaluate the impact of description on product appeal.
prompt:
Analyze the attractiveness and completeness of product descriptions based on product description and number of reviews .This is presented in a scatter plot: the horizontal coordinate is the length of the product description and the vertical coordinate is the number of reviews.
AI Chart Interpretation
The scatter plot illustrates the relationship between the length of product descriptions and the number of reviews on Amazon. From the chart, we can observe that:
Most products have relatively short descriptions (less than 500 characters) and a varying number of reviews. There is a noticeable clustering of products with shorter descriptions and a smaller number of reviews.
There is no clear correlation between longer descriptions and a higher number of reviews, as products with longer descriptions do not consistently have more reviews. In fact, there are many instances of products with extended descriptions that have few reviews, indicating that a longer description might not necessarily enhance review count.
Correlation heat map
We can use correlation heatmaps to see the correlation relationship between data.
prompt:
Reads the value of the column. Draw a heat map to show the correlation between these values.
