Spaces:
Running
Running
| # with cluster profiles | |
| import gradio as gr | |
| import pickle | |
| import pandas as pd | |
| import shap | |
| import matplotlib.pyplot as plt | |
| # Load model | |
| filename = 'xgb_h_new.pkl' | |
| with open(filename, 'rb') as f: | |
| loaded_model = pickle.load(f) | |
| # Setup SHAP | |
| explainer = shap.Explainer(loaded_model) | |
| # Employee Profiles (From SPSS 3-Cluster Solution) | |
| employee_profiles = { | |
| "🏆 Leslie Knope": [4.716, 4.792, 4.864, 4.588, 4.849, 4.601], # Cluster group 1 averages - high engagement, strong support, high workload | |
| "⚠️ Kevin Malone": [3.045, 3.122, 3.129, 2.886, 3.113, 2.197], # Cluster group 2 averages - disengaged, low recognition, weak support | |
| "🌱 Jim Halpert": [3.885, 3.992, 4.119, 3.704, 4.090, 3.377] # Cluster group 3 averages - Moderately engaged, could be more recognized - room to grow | |
| } | |
| # Define the prediction function | |
| def main_func(WellBeing, SupportiveGM, Engagement, Workload, WorkEnvironment, Merit): | |
| new_row = pd.DataFrame({ | |
| 'WellBeing': [WellBeing], | |
| 'SupportiveGM': [SupportiveGM], | |
| 'Engagement': [Engagement], | |
| 'Workload': [Workload], | |
| 'WorkEnvironment': [WorkEnvironment], | |
| 'Merit': [Merit] | |
| }) | |
| # Predict probability | |
| prob = loaded_model.predict_proba(new_row) | |
| shap_values = explainer(new_row) | |
| # Calculate probability values | |
| stay_prob = round((1 - float(prob[0][0])) * 100, 2) | |
| leave_prob = round(float(prob[0][0]) * 100, 2) | |
| # Dynamic risk label: Changes color & text based on probability | |
| risk_label = "🔴 High Risk of Turnover" if leave_prob > 50 else "🟢 Low Risk of Turnover" | |
| risk_color = "red" if leave_prob > 50 else "green" | |
| risk_html = f""" | |
| <div style='padding: 15px; border-radius: 8px;'> | |
| <span style='color: {risk_color}; font-size: 26px; font-weight: bold;'>{risk_label}</span> | |
| <ul style='list-style-type: none; padding-left: 0; font-size: 20px; font-weight: bold; color: #0057B8;'> | |
| <li>🧲 Likelihood of Staying: {stay_prob}%</li> | |
| <li>🚪 Likelihood of Leaving: {leave_prob}%</li> | |
| </ul> | |
| </div> | |
| """ | |
| # Key Insights (Updated for 0.1-point increments) | |
| insights_html = "<div style='font-size: 18px;'>" | |
| for feature, shap_val in dict(zip(new_row.columns, shap_values.values[0])).items(): | |
| impact = round(shap_val * 10, 2) # Scaling impact for 0.1 changes | |
| icon = "📈" if shap_val > 0 else "📉" | |
| effect = "raises turnover risk" if shap_val > 0 else "improves retention" | |
| insights_html += f"<p style='margin: 5px 0;'> {icon} <b>Each 0.1-point increase in {feature} {effect} by {abs(impact)}%.</b></p>" | |
| insights_html += "</div>" | |
| # Final Layout (Risk + Key Insights) | |
| final_layout = f""" | |
| <table style='width:100%; border-collapse: collapse; margin-top: 10px; background-color: #FFFFFF;'> | |
| <tr> | |
| <td style='width: 33%; padding: 15px; background-color: #FFFFFF; border-radius: 8px; vertical-align: top;'> | |
| {risk_html} | |
| </td> | |
| <td style='width: 67%; padding: 15px; background-color: #FFFFFF; border-radius: 8px; vertical-align: top;'> | |
| <b style='color: #0057B8; font-size: 22px;'>Key Insights:</b> | |
| {insights_html} | |
| </td> | |
| </tr> | |
| </table> | |
| """ | |
| # Retention vs. Turnover Chart | |
| fig, ax = plt.subplots() | |
| categories = ["Stay", "Leave"] | |
| values = [stay_prob, leave_prob] | |
| colors = ["#0057B8", "#D43F00"] | |
| ax.barh(categories, values, color=colors) | |
| for i, v in enumerate(values): | |
| ax.text(v + 2, i, f"{v:.2f}%", va='center', fontweight='bold', fontsize=12) | |
| ax.set_xlabel("Probability (%)") | |
| ax.set_title("Retention vs. Turnover Probability") | |
| plt.tight_layout() | |
| prob_chart_path = "prob_chart.png" | |
| plt.savefig(prob_chart_path, transparent=True) | |
| plt.close() | |
| # SHAP Chart | |
| fig, ax = plt.subplots() | |
| shap.plots.bar(shap_values[0], max_display=6, show=False) | |
| ax.set_title("Key Drivers of Turnover Risk") | |
| plt.tight_layout() | |
| shap_plot_path = "shap_plot.png" | |
| plt.savefig(shap_plot_path, transparent=True) | |
| plt.close() | |
| return final_layout, prob_chart_path, shap_plot_path | |
| # UI Setup | |
| with gr.Blocks() as demo: | |
| gr.Markdown(""" | |
| <div style="display: flex; justify-content: center; align-items: center;"> | |
| <img src="https://cdn-uploads.huggingface.co/production/uploads/67a2a1b7cfdd6c9f7fa6359f/b1eeV4z6_mLNUGGRa4Pb9.png" | |
| alt="Hilton Logo" width="250px"> | |
| </div> | |
| """) | |
| gr.Markdown("<h1 style='color: #0057B8;'>Hilton Team Member Retention Predictor</h1>") | |
| gr.Markdown(""" | |
| <div style='font-size: 20px; color: #0057B8;'> | |
| ✨ <b>Welcome to Hilton’s Employee Retention Predictor</b><br> | |
| This tool helps <b>Sales & Marketing leaders and Front Office Operations teams</b>—The Face of Hilton— | |
| assess <b>team member engagement</b> and predict <b>turnover risk</b> using AI-powered insights.<br> | |
| 🔍 <b>Understand what drives retention and make data-driven decisions to keep top talent.</b> | |
| </div> | |
| # Dropdown for Employee Profiles | |
| profile_dropdown = gr.Dropdown(choices=list(employee_profiles.keys()), label="Select Employee Profile") | |
| # Sliders for input features | |
| with gr.Row(): | |
| WellBeing = gr.Slider(label="WellBeing Score", minimum=1, maximum=5, value=4, step=0.1) | |
| SupportiveGM = gr.Slider(label="Supportive GM Score", minimum=1, maximum=5, value=4, step=0.1) | |
| Engagement = gr.Slider(label="Engagement Score", minimum=1, maximum=5, value=4, step=0.1) | |
| with gr.Row(): | |
| Workload = gr.Slider(label="Workload Score", minimum=1, maximum=5, value=4, step=0.1) | |
| WorkEnvironment = gr.Slider(label="Work Environment Score", minimum=1, maximum=5, value=4, step=0.1) | |
| Merit = gr.Slider(label="Merit Score", minimum=1, maximum=5, value=4, step=0.1) | |
| submit_btn = gr.Button("🔎 Click Here to Analyze Retention") | |
| # Output elements | |
| prediction = gr.HTML() | |
| with gr.Row(): | |
| prob_chart = gr.Image(label="Retention vs. Turnover Probability", type="filepath") | |
| shap_plot = gr.Image(label="Key Drivers of Turnover Risk", type="filepath") | |
| # Allow profile selection to update sliders | |
| def update_sliders(profile): | |
| if profile in employee_profiles: | |
| return employee_profiles[profile] | |
| return [4, 4, 4, 4, 4, 4] | |
| profile_dropdown.change(update_sliders, inputs=[profile_dropdown], outputs=[WellBeing, SupportiveGM, Engagement, Workload, WorkEnvironment, Merit]) | |
| submit_btn.click(main_func, [WellBeing, SupportiveGM, Engagement, Workload, WorkEnvironment, Merit], [prediction, prob_chart, shap_plot]) | |
| demo.launch() | |