An early-stage e-commerce company, offering a wide range of products from daily essentials to high-end electronics. Despite a high volume of traffic, the conversion rate from browsing to purchasing remains low. We, therefore, sought to predict the purchase probability of each platform user and implement a targeted discount strategy that moves them farther along the buyer journey.
- Build a model that predicts the purchase propensity of each user, enabling targeted discount campaigns.
- RFM Analysis: Conducted a detailed RFM (Recency, Frequency, Monetary) analysis to segment users based on their purchase behaviour.
- Propensity Modeling: Used propensity modelling to forecast user behaviour and identify users who require incentives to make a purchase.
The dataset provided by the company includes the following features:
User_id: Unique identifier for each user.Session_id: A unique identifier generated every time a user visits the platform.DateTime: Timestamp indicating when an action is performed by the user.Category: The category of the product being interacted with.SubCategory: The subcategory of the product.Action: The type of action performed by the user, such as viewing a product, reading reviews, making a purchase, adding a product to the cart, etc.Quantity: The number of products ordered in a transaction.Rate: The price of a single product unit.Total Price: The total order price, calculated as Quantity multiplied by Rate.
- Exploratory Data Analysis (EDA): Initial analysis to understand the data structure, distribution, and patterns.
- Feature Engineering: Developed relevant features that capture user behaviour and purchase patterns.
- RFM Analysis: Conducted customer segmentation based on Recency, Frequency, and Monetary values to understand customer value.
- Propensity Modeling: Applied statistical models to predict the likelihood of a user making a purchase based on their behaviour and characteristics.