app.py

import streamlit as st
import pickle
import pandas as pd
from catboost import CatBoostClassifier

# Load the trained model and unique values from the pickle file
with open('model_and_key_components.pkl', 'rb') as file:
    saved_components = pickle.load(file)

model = saved_components['model']
unique_values = saved_components['unique_values']

# Define the Streamlit app
def main():
    st.title("Employee Attrition Prediction App 🕵️‍♂️")
    st.sidebar.title("Model Settings ⚙️")

    # Sidebar inputs
    with st.sidebar.expander("View Unique Values 🔍"):
        st.write("Unique values for each feature:")
        for column, values in unique_values.items():
            st.write(f"- {column}: {values}")

    # Main content
    st.write("Welcome to the Employee Attrition Prediction App! 🚀")
    st.write("This app helps HR practitioners predict employee attrition using a trained CatBoost model.")
    st.write("Please provide the following information to make a prediction:")

    # Define layout with three columns
    col1, col2, col3 = st.columns(3)

    # Column 1
    with col1:
        age = st.number_input("Age", min_value=18, max_value=70)
        monthly_income = st.number_input("Monthly Income")
        num_companies_worked = st.number_input("Number of Companies Worked")
        percent_salary_hike = st.number_input("Percent Salary Hike", min_value=0, max_value=25)
        training_times_last_year = st.number_input("Training Times Last Year", min_value=0, max_value=6)

    # Column 2
    with col2:
        department = st.selectbox("Department", ['Sales', 'Research & Development', 'Human Resources'])
        environment_satisfaction = st.selectbox("Environment Satisfaction", [1, 2, 3, 4])
        job_role = st.selectbox("Job Role", ['Sales Executive', 'Research Scientist', 'Laboratory Technician',
                                              'Manufacturing Director', 'Healthcare Representative', 'Manager',
                                              'Sales Representative', 'Research Director', 'Human Resources'])
        job_satisfaction = st.selectbox("Job Satisfaction", [1, 2, 3, 4])
        work_life_balance = st.selectbox("Work Life Balance", [1, 2, 3, 4])

    # Column 3
    with col3:
        over_time = st.checkbox("Over Time")
        relationship_satisfaction = st.selectbox("Relationship Satisfaction", [1, 2, 3, 4])
        years_since_last_promotion = st.number_input("Years Since Last Promotion")
        years_with_curr_manager = st.number_input("Years With Current Manager")

    # Predict button
    if st.button("Predict 📊"):

        # Create a DataFrame to hold the user input data
        input_data = pd.DataFrame({
            'Age': [age],
            'Department': [department],
            'EnvironmentSatisfaction': [environment_satisfaction],
            'JobRole': [job_role],
            'JobSatisfaction': [job_satisfaction],
            'MonthlyIncome': [monthly_income],
            'NumCompaniesWorked': [num_companies_worked],
            'OverTime': [over_time],
            'PercentSalaryHike': [percent_salary_hike],
            'RelationshipSatisfaction': [relationship_satisfaction],
            'TrainingTimesLastYear': [training_times_last_year],
            'WorkLifeBalance': [work_life_balance],
            'YearsSinceLastPromotion': [years_since_last_promotion],
            'YearsWithCurrManager': [years_with_curr_manager]
        })

        # Reorder columns to match the expected order
        input_data = input_data[['Age', 'Department', 'EnvironmentSatisfaction', 'JobRole', 'JobSatisfaction',
                                 'MonthlyIncome', 'NumCompaniesWorked', 'OverTime', 'PercentSalaryHike',
                                 'RelationshipSatisfaction', 'TrainingTimesLastYear', 'WorkLifeBalance',
                                 'YearsSinceLastPromotion', 'YearsWithCurrManager']]

        # Make predictions
        prediction = model.predict(input_data)
        probability = model.predict_proba(input_data)[:, 1]
        
        # Display prediction probability
        if prediction[0] == 1:
            st.subheader("Prediction Probability 📈")
            st.write(f"The probability of the employee leaving is: {probability[0]*100:.2f}%")
        
        # Display characteristic-based recommendations
        st.subheader("Recommendations for Retaining The Employee 💡:")
        if job_satisfaction == 1 or environment_satisfaction == 1:
            st.markdown("- **Job and Environment Satisfaction**: Enhance job and environment satisfaction through initiatives such as recognition programs and improving workplace conditions.")
        if years_since_last_promotion > 5:
            st.markdown("- **Opportunities for Advancement**: Implement a transparent promotion policy and provide opportunities for career advancement.")
        if years_with_curr_manager > 5:
            st.markdown("- **Change in Reporting Structure**: Offer opportunities for a change in reporting structure to prevent stagnation and promote growth.")
        if percent_salary_hike < 5:
            st.markdown("- **Salary and Benefits Adjustment**: Consider adjusting salary and benefits packages to remain competitive and reward employee loyalty.")
        if training_times_last_year < 2:
            st.markdown("- **Employee Development**: Invest in employee development through training programs and continuous learning opportunities.")
        if over_time:
            st.markdown("- **Workload Evaluation**: Evaluate workload distribution and consider implementing measures to prevent overwork, such as workload balancing and flexible scheduling.")
        if relationship_satisfaction == 1:
            st.markdown("- **Positive Work Environment**: Foster positive relationships and a supportive work environment through team-building activities and open communication channels.")
        if monthly_income < 5000:
            st.markdown("- **Compensation Review**: Review compensation structures and adjust salaries to align with industry standards and employee expectations.")
        if num_companies_worked > 5:
            st.markdown("- **Address High Turnover**: Identify reasons for high turnover and address issues related to job stability, career progression, and organizational culture.")
        if work_life_balance == 1:
            st.markdown("- **Work-Life Balance Initiatives**: Promote work-life balance initiatives, such as flexible work arrangements and wellness programs, to support employee well-being.")

        # General recommendation for all negative predictions
        st.markdown("- **Exit Interviews**: Conduct exit interviews to gather feedback and identify areas for improvement in retention strategies.")

if __name__ == "__main__":
    main()