Machine Learning Pipeline

Attiriton Rate Prediction

Name : Lau Sook Han Gayle Date: 31/10/22 Email: laugayle@gmail.com

Data Overview

Independent Features

Attribute Description

• Age - Age of the member when signing up as a member
• Gender - Gender of the member
• Monthly Income - Monthly declared income of the member in SGD
• Travel Time - Estimated amount of time needed to travel to the club from home (mins)
• Qualification - Education qualification level of the member (1-Diploma, 2-Bachelors, 3-Masters, 4-PH.D)
• Work Domain - Work domain of the member
• Usage rate - Average number of days in a week visiting the country club
• Branch - Location of the branch that the member visits
• Membership - Membership tier (1-Normal 2-Bronze 3-Silver 4-Gold)
• Months - Number of months as a member of the country club
• Birth Year - Year the member was born
• Usage Time - Average number of hours spent in the country club per visit
• Usage - Usage Time * Usage Rate (new)

Target Feature

• Target Feature: Attrition (If the member left: 0 = No, 1 = Yes)

Unique Identifying Features

• Member Unique ID: Unique member ID (removed)

Catergorical Features

• Categorical features: Qualification, Work Domain, Branch, Membership

Binary Features

• Binary features: Gender, Attrition

Numerical Features

• Numerical Features: Age,Travel Time,Monthly Income, Usage Rate, Months, Birth Year, Usage Time

Sypnopsis of the problem.

• Classification: predict member attrition using the provided dataset to help a country club to formulate policies to reduce attrition. In your submission, you are to evaluate at least 3 suitable models for predicting member attrition.

Overview of Submitted folder

. ├── eda.ipnyb ├── data │ └── score.db # removed ├── results │ └── testing accuracy.csv ├── requirements.txt ├── run.sh └── src ├── └── run.py

Executing the pipeline

**run.py

Steps:

1. Imports the data from .db file

2. Data Cleaning (data is cleaned - 'Age', 'Monthly Income', Birth Year, 'Qualification', 'Travel Time)'

3. Feature Engineering (one-hot encoding/ordinal encoding for categorical data)

4. Data split into training and testing data (80/20)

5. Building of the Models

• Logistic Regression
• Decision Tree Model
• Boosting Decision Tree
• Bagging Decision Tree
• Random Forest Model
• Support Vector Machine
• KNN
• Naive Bayes

6. Results

Accuracy and Recall

Running of machine learning pipeline.

Machine learning model created with python 3 and bash script.

Installing dependencies

Paste the following command on your bash terminal to download dependencies

pip install -r requirements.txt

Running the Machine Learning Pipeline

Past the followin command on your bash terminal to grant permission to execute the 'run.sh' file

chmod +x run.sh

Paste the following command on the bash terminal to run the machine learning programme

./run.sh