Update app.py

app.py

@@ -5,6 +5,8 @@ import pandas as pd
 import joblib
 import os
 import warnings
+import shap
+import matplotlib.pyplot as plt
 
 # Suppress XGBoost warnings
 warnings.filterwarnings("ignore", category=UserWarning, message=".*WARNING.*")
@@ -23,10 +25,10 @@ def load_model():
 
 model = load_model()
 
-# Prediction function
+# Prediction function with dynamic threshold
 def predict_employee_status(satisfaction_level, last_evaluation, number_project,
                             average_monthly_hours, time_spent_company, 
-                            work_accident, promotion_last_5years, salary, department):
+                            work_accident, promotion_last_5years, salary, department, threshold=0.5):
     
     # One-hot encode the department
     departments = [
@@ -43,7 +45,7 @@ def predict_employee_status(satisfaction_level, last_evaluation, number_project,
         "last_evaluation": [last_evaluation],
         "number_project": [number_project],
         "average_monthly_hours": [average_monthly_hours],
-        "time_spend_company": [time_spent_company],
+        "time_spent_company": [time_spent_company],
         "Work_accident": [work_accident],
         "promotion_last_5years": [promotion_last_5years],
         "salary": [salary],
@@ -60,33 +62,54 @@ def predict_employee_status(satisfaction_level, last_evaluation, number_project,
         dmatrix = xgb.DMatrix(input_df)
         prediction = model.predict(dmatrix)
         prediction_prob = prediction[0]
-        result = "✅ Employee is likely to quit." if prediction_prob > 0.3 else "✅ Employee is likely to stay."
-        return f"{result}"
+
+        # Apply the dynamic threshold
+        result = "✅ Employee is likely to quit." if prediction_prob >= threshold else "✅ Employee is likely to stay."
+        explanation = explain_prediction(input_df)
+        return f"{result} (Probability: {prediction_prob:.2%})\n\nExplanation:\n{explanation}"
     except Exception as e:
         return f"❌ Error: {str(e)}"
 
-# Gradio interface
-interface = gr.Interface(
-    fn=predict_employee_status,
-    inputs=[
-        gr.Number(label="Satisfaction Level (0.0 - 1.0)"),
-        gr.Number(label="Last Evaluation (0.0 - 1.0)"),
-        gr.Number(label="Number of Projects (1 - 10)"),
-        gr.Number(label="Average Monthly Hours (80 - 320)"),
-        gr.Number(label="Time Spent at Company (Years)"),
-        gr.Radio([0, 1], label="Work Accident (0 = No, 1 = Yes)"),
-        gr.Radio([0, 1], label="Promotion in Last 5 Years (0 = No, 1 = Yes)"),
-        gr.Radio([0, 1, 2], label="Salary (0 = Low, 1 = Medium, 2 = High)"),
-        gr.Dropdown(
-            ['RandD', 'accounting', 'hr', 'management', 'marketing',
-             'product_mng', 'sales', 'support', 'technical'], 
-            label="Department"
-        )
-    ],
-    outputs="text",
-    title="Employee Retention Prediction System",
-    description="Predict whether an employee is likely to stay or quit based on their profile.",
-    theme="dark"
-)
+# SHAP Explainability
+def explain_prediction(input_df):
+    explainer = shap.TreeExplainer(model)
+    shap_values = explainer.shap_values(input_df)
+    
+    # Generating SHAP explanation for this prediction
+    shap.initjs()
+    plt.figure()
+    shap.waterfall_plot(shap.Explanation(values=shap_values[0], 
+                                         base_values=explainer.expected_value,
+                                         data=input_df.iloc[0].values,
+                                         feature_names=input_df.columns))
+    plt.savefig("shap_explanation.png")
+    return "SHAP explanation generated for this prediction."
+
+# Gradio interface with dynamic threshold
+def gradio_interface():
+    interface = gr.Interface(
+        fn=predict_employee_status,
+        inputs=[
+            gr.Number(label="Satisfaction Level (0.0 - 1.0)"),
+            gr.Number(label="Last Evaluation (0.0 - 1.0)"),
+            gr.Number(label="Number of Projects (1 - 10)"),
+            gr.Number(label="Average Monthly Hours (80 - 320)"),
+            gr.Number(label="Time Spent at Company (Years)"),
+            gr.Radio([0, 1], label="Work Accident (0 = No, 1 = Yes)"),
+            gr.Radio([0, 1], label="Promotion in Last 5 Years (0 = No, 1 = Yes)"),
+            gr.Radio([0, 1, 2], label="Salary (0 = Low, 1 = Medium, 2 = High)"),
+            gr.Dropdown(
+                ['RandD', 'accounting', 'hr', 'management', 'marketing',
+                 'product_mng', 'sales', 'support', 'technical'], 
+                label="Department"
+            ),
+            gr.Slider(0.1, 0.9, value=0.5, step=0.05, label="Prediction Threshold")
+        ],
+        outputs="text",
+        title="Employee Retention Prediction System (With SHAP Explainability)",
+        description="Predict whether an employee is likely to stay or quit based on their profile. Adjust the threshold for accurate predictions.",
+        theme="dark"
+    )
+    interface.launch()
 
-interface.launch()
+gradio_interface()