E_commerce

Data source

the file contains the dataset pertaining to 10K records of information about:

Date of the order and the shipment
Shipment Mode
Customer Id,name,segment
Adress of the customer (country,state,city,postel code)
Product Id,name, category,sub-category
Sales
Discount
Profit
Quantity

Project Introduction

The project focused on exploratory data analysis and database design (SQL), identification of data-driven business strategies and their presentation in an interactive Tableau dashboard.

Data base design

On the firs step, I created a data_base using SQLite &Python and made a ERD. For further analysis I performed some queries.

Explaroty data analysis

In EDA I investigated:

states which have made the highest number of orders and highest revenue and profit;
customer buying pattern depending on the time of year;
the worst and the best sub-categories:
- in terms of profit margin and sales;
- in terms of profit margin and provided discount
typical order overview:
- The most frequently purchased product sub-categories together;
- Orders destribution according to number of categories in the order;
- sub-categories by delivery mode
correlation between discount and sales

Additionaly, to understand the customers buying pattern 3 parameters were calculated :

Monetary
Frequency
Recency (how many days has passed from the last order date in the dataset to customers last purchase)

ML Clustering with K-means

To find meaningful structure, explanatory underlying processes, generative features, and groupings inherent in a set of examples I used unsupervised learning method - Clustering with K-means.

After removing outlires, feature scaling I used The Elbow Method and Silhouette Analysis to determine the optimal number of clusters into which the data may be clustered.

After obtaining the optimal number of clusters (4), I used the K-means clustering algorithm to divide the clients into 4 clusters according to Monetary,Frequency,Recency.