Employee_Retention_Predictive_Modeling

Executive Summary:

The HR department of Salifort Motors want to improve employee satisfaction within the company. They have collected data and want to know what causes employees to leave the company. Additionally, they want to predict which employees will leave the company in the future. After some analysis, it is clear that the last evaluation date of an employee, the number of projects an employee has, employee tenure and the number of hours a employee works have the highest impact on if a employee will leave or not. With this information a Random Forest model was created to predict future churn.

With this information, the HR department can implement programs that will help improve factors that lead to employees leaving.

Business Problem:

This project addresses the issue of employee rentention at Salifort Motors. The goal of the following analysis is to understand what factors cause employees to leave and to create a model that predicts whether or not a employee will leave the company.

Methodology

Since the variable we are analyzing ("left") is a binary catergorical value, a logistic regression model and tree-based models were built. The Random Forest model's predicitive power outperformed logisitc regression and decision tree models.

Additionally, many visuals were created to analyze the relationship between variables.

Skills:

- Python: pandas, numpy, seaborn, matplotlib, xgboost,sklearn

Key Insights & Recommendation:

The top 4 variables that help predict wheater a employee will leave or not are:

1. last_evaluation
2. number_project
3. tenure
4. overworked

After various correlation tests stakeholders should consider:

• Placing a cap on the number or projects and hours individual employees can work
• Implementing a more frequent promotion program based on tenure
• Implementing a reward system for working overtime
• Hiring more employees to decrease workload.