irwinhentriansa / KnowingYourCustomer_RFM-Clustering-Recommender_Final_Project

What kind of products that cluster and customer segment need of based on time series to get an insight.

JCDSJKT09 -Final Project-

---

Final Project as a requirement for completing the Job Connector Data Science in Purwadhika Startup & Coding School Jakarta Batch 09.

RFM is a method used for analyzing customer value

Original Data Set taken from : https://www.kaggle.com/mashlyn/online-retail-ii-uci

This dataset have 8 columns and 1067371 rows

---

The Brief Background

You've got to start with the customer experience and work back toward the technology - not the other way around.
-Steve Jobs.

Business Insights.

• Does it matter knowing what kind of products that they love ?

• When the best time to give an emphaty tou your customer ?

• How do you get in touch with your Customer?

Workflow

Let's Take a Look The Features:

• InvoiceNo: Invoice number. Nominal. A 6-digit integral number uniquely assigned to each transaction. If this code starts with the letter 'c', it indicates a cancellation.
• StockCode: Product (item) code. Nominal. A 5-digit integral number uniquely assigned to each distinct product.
• Description: Product (item) name. Nominal.
• Quantity: The quantities of each product (item) per transaction. Numeric.
• InvoiceDate: Invice date and time. Numeric. The day and time when a transaction was generated.
• UnitPrice: Unit price. Numeric. Product price per unit in poundsterling.
• CustomerID: Customer number. Nominal. A 5-digit integral number uniquely assigned to each customer.
• Country: Country name. Nominal. The name of the country where a customer resides.

---

1. Data Cleaning and Preprocessing

A. Cleaning

• Checking The Null Values and checking the relatable data through each rows using msno

• Drop The Duplicated Values, and keep the first rows where data has represented by duplicates.

• Remove the minus values on Price and Quantity

• Imputing missing values on CustomerID which have been overlap checking in Invoice

• Imputing missing values on Description which have been overlaping with StockCode

B. Enhancing Features to Get Time Series Features/Columns

• We have an InvoiceDate disparting its features to get a month, year, hour, days, and week.

2. Exploratory Data Analysis (EDA)

Get an insights from the data where have a story behind.

3. RFM Analysis and Clustering

Those clustering based on goal oriented by Elbow-Methods, Silhoutte Score, DaviesBouldin Score then we have 5 clustering are representing by distribution of labelling to each customer.

4. Product Recommender and Bundling through time based on clustering and RFM

Recommender System

Product Bundling

Conclusion

The best marketing doesn't feel like marketing - Tom Fishburne