The dataset consists of dummy employee details of a company. It includes employees' education details, joining year, city of office, payment tier, age, gender, if the employee is ever kept out of projects for 1 month or more and experience in current field. The task is to predict whether the employee will leave the company in the next 2 years.

For the project, first I have done EDA and feature selection. Then, I have trained three models: LogisticRegression, RandomForestClassifier and XGBClassifier.

I have created a Pipeline for each of these models, such that the data transformation and model training/predictions steps are assembled together.

• Python
• Scikit-learn
• Machine Learning Pipeline
• Docker
• Streamlit

• Clone the project repo and open it.

• Create a virtual environment for the project using

  pipenv shell

• Install required packages using

  pipenv install

• Build the docker image using

  sudo docker build -t employee_future .

• Run the docker container using

  sudo docker run -p 5000:5000 employee_future

• Open the URL http://localhost:5000/ to run and test the app.

• Open the Deploy an app page of Streamlit.
• Enter the GitHub repository details in which the streamlit_app.py file and model binaries are stored.
• Click on Deploy button.
• Open the URL employeefutureprediction.streamlit.app to run and test the app.

Selected Model (RandomForestClassifier) Test Set Accuracy = 86.57 %