/E-Commerce-Analytics-Project

E-Commerce Analytics with RFM Approach

Primary LanguageJupyter Notebook

E-Commerce Analytics

( The RFM Approach )

ABSTRACT :

Analyze and segment the customers of an e-commerce company by using the RFM approach.
This will enable the e-commerce company to optimize their retention and acquisition strategies.

Market Outlook:

E-commerce stores which became success stories were successful in targeting the desired customers.
One of the techniques by which they were able to achieve this was customer segmentation i.e. by segmenting the existing customers based on frequency of purchases, monetary value etc.

E-commerce stores who designed market strategies based on mass marketing soon realized the need of customer segmentation as an alternative to save cost and efforts in the digital sphere.

In a real-world segmentation scenario, there might be hundreds of variables which can be used but broadly they segment the customers by the following characteristics:

  • Geographic - Segments based on country, state, and city.

  • Demographic - Segments based on gender, age, income, education level, etc.

  • Psychographic - Segments based on geography, lifestyle, age and religious beliefs, etc.

  • Behavior - Segments based on consumer personality traits, attitudes, interests, and lifestyles.

Overview of the problem :

  • Data File Provided: A single file is provided which contains data related to the ecommerce transactions.
  • Contents:
    Date-time of sale, Customer shipping location, Price of single unit from 2016 to 2017.

Data and Problem Detail :

  • Draw meaningful insights from 2 years of data & provide brief details based on the monetary value, frequency of buy, etc.

Objective :

  • Build an unsupervised learning model which can enable your company
    to analyze their customers via RFM (Recency, Frequency and Monetary value) approach.

Steps to be followed:

attachment:image-3.png

Libraries Used :

  • NUMPY, PANDAS, Matplotlib, Warnings

Data Cleaning & Preprocessing:

  • Checking Null, Unique and Duplicate Values
  • Drop unneccessary/ duplicate value columns after checking with supervisor
  • Drop missing value column
  • Format column as required (Date)
  • Feature Extraction using column formatting and creating function

Data Visualization :

EDA - Descriptive Analysis

  • The Recency v/s Density Plot

attachment:recency.png

  • The Monetary v/s Density Plot

attachment:monetary.png

K- Means Cluster:

  • Here 4 clusters are used. They are,
                             'Platinum'
                             'Gold'
                             'Silver'
                             'Bronze'
  • Plot of Elbow Method for Optimal K Value

attachment:Kmeans%20cluster.png