Comparing to Find The Best Regression Model for Predicting

This notebook will show you step by step to build machine learning to predict compensation of some employee. I've got the dataset from official website of https://www.phoenixopendata.com, you can explore this website for more dataset.

Step by Step:

1. Data Understanding: Before we start to process our data we MUST understand a feautures of the data

Feautures

• Name = Full name of the employee.
• Benefits Category = Benefit category of employee.
• Department = Employee department.
• Job Title = Job title of employee.
• Full/Part Time = Employee full or part-time status.
• Hire Date = Employee hire date.
• Termination Date = Employee termination date (if applicable).
• Hourly Rate = Employee hourly pay rate.
• Regular Pay = Annualized base pay of employee.
• Oveprtime Pay = Employee overtime paid.
• Other Pay = Other Pay includes productivity enhancement pay, shift differential pay, standby pay, allowances, leave payout, and incentives

2. Data Preprocessing: In this step we will processing our raw data using several techniques like eliminated, manipulated,etc.

Preprocess:

• Clear rows with with negative values
• Convert Hire Date & Termination Date from object into datetime
• Add new column Experience (Termination Date – Hire Date)
• Drop column Hire Date & Termination Date
• Encoding Full/Part Time (‘F’ = 1 , ‘P’ = 0)
• Checking and Handling Missing Values
• Checking and Handling Outliers

Checking and Handling missing values and outliers is IMPORTANT step when we want to build machine learning with Linear Regression, cuz this model is very sensitive with outliers.

3. Build and Find the best model for our Machine Learning: We will compare several Regression Algorithm and choose the right model that we can use to our Machine Learning.

Linear Regression:

• Standard Linear Regression (Baseline)
• Ridge Regression
• Lasso Regression

Non-Linear Regression

• Decicision Tree
• XGBOOST
• Random Forest