https://github.com/chhavi-trivedi/tensorflowbusinessex

A machine learning algorithm based on our available data that can predict if a customer will buy again from the Audiobook company.
https://github.com/chhavi-trivedi/tensorflowbusinessex

machine-learning-algorithms tensorflow-examples

Last synced: 15 days ago
JSON representation

A machine learning algorithm based on our available data that can predict if a customer will buy again from the Audiobook company.

• Host: GitHub
• URL: https://github.com/chhavi-trivedi/tensorflowbusinessex
• Owner: Chhavi-Trivedi
• Created: 2020-10-13T09:31:33.000Z (about 4 years ago)
• Default Branch: main
• Last Pushed: 2020-12-17T06:40:29.000Z (almost 4 years ago)
• Last Synced: 2024-10-17T13:14:11.917Z (about 1 month ago)
• Topics: machine-learning-algorithms, tensorflow-examples
• Language: Jupyter Notebook
• Homepage:
• Size: 14.6 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Practical example. Audiobooks

### Problem

You are given data from an Audiobook app. Logically, it relates only to the audio versions of books. Each customer in the database has made a purchase at least once, that's why he/she is in the database. We want to create a machine learning algorithm based on our available data that can predict if a customer will buy again from the Audiobook company.

The main idea is that if a customer has a low probability of coming back, there is no reason to spend any money on advertizing to him/her. If we can focus our efforts ONLY on customers that are likely to convert again, we can make great savings. Moreover, this model can identify the most important metrics for a customer to come back again. Identifying new customers creates value and growth opportunities.

You have a .csv summarizing the data. There are several variables: Customer ID, Book length in mins_avg (average of all purchases), Book length in minutes_sum (sum of all purchases), Price Paid_avg (average of all purchases), Price paid_sum (sum of all purchases), Review (a Boolean variable), Review (out of 10), Total minutes listened, Completion (from 0 to 1), Support requests (number), and Last visited minus purchase date (in days).

So these are the inputs (excluding customer ID, as it is completely arbitrary. It's more like a name, than a number).

The targets are a Boolean variable (so 0, or 1). We are taking a period of 2 years in our inputs, and the next 6 months as targets. So, in fact, we are predicting if: based on the last 2 years of activity and engagement, a customer will convert in the next 6 months. 6 months sounds like a reasonable time. If they don't convert after 6 months, chances are they've gone to a competitor or didn't like the Audiobook way of digesting information.

The task is simple: create a machine learning algorithm, which is able to predict if a customer will buy again.

This is a classification problem with two classes: won't buy and will buy, represented by 0s and 1s.