Ecommerce Sales Data Transformation

Client Background: An ecommerce company struggling with unorganized sales data.

Objective: This project aimed for two key objectives- Customer Segmentation, focusing on targeted marketing and improved customer satisfaction, and Customer Categories, aimed at strategic decision-making, personalized campaigns, and long-term customer loyalty. The dual goals underscored our commitment to refining strategies and making customer-centric decisions for sustained business growth.

In the initial phase of our data analytics project, we diligently collected data from an E-commerce database encompassing the purchase records of around 4000 customers over a year. This involved gathering detailed customer profiles, purchase histories, and transaction timestamps. Subsequently, our focus shifted to Data Preparation, where we meticulously cleaned the dataset to address inconsistencies and missing values, ensuring its accuracy. Additionally, normalization and standardization techniques were applied to promote uniformity for effective variable comparison. This thorough Data Collection and Preparation phase laid a robust foundation, enhancing the reliability of subsequent analyses and predictive model development by providing a clean and accurate dataset.

Countries: Analyzed geographical distribution of sales data to identify key markets and

potential areas for growth.

Customers and Products: Explored customer and product-related variables to understand

their impact on sales.

Cancelling Orders - Investigated the reasons behind order cancellations to enhance

customer satisfaction.

Stock Code - Examined stock codes for consistency and accuracy in product

identification.

Basket Price - Studied basket prices to identify trends in purchasing behavior.

Product Description: Analyzed product descriptions to gain insights into product features

and characteristics.

Data Encoding - Implemented encoding techniques to categorize products for efficient

analysis.

Clusters of Product - Utilized clustering algorithms to group products based on

similarities.

Characterizing the content of Clusters - Examined the characteristics of product

clusters for targeted marketing strategies.

Grouping Product - Grouped products to analyze customer preferences and purchasing

patterns.

Time Splitting of the Dataset - Split the dataset based on time to identify temporal

trends in customer behavior.

Grouping Orders - Grouped orders for a holistic view of customer transactions.

Creating Customer Categories:

Data Encoding - Applied data encoding techniques to categorize customers based on

their preferences.

Creating Categories - Established customer categories to tailor marketing strategies and

promotions.

Gradient Boosting Classifier: Gradient Boosting Classifier Chosen for

superior performance (75.23% precision).