Upload app.py

app.py

@@ -4,17 +4,25 @@ import pandas as pd
 import ast
 import numpy as np
 import os
+import matplotlib.pyplot as plt
 
 # Set the option to opt into future behavior
 pd.set_option('future.no_silent_downcasting', True)
 
 # List of options for the dropdown
-workclass_options = sorted(['State-gov', 'Self-emp-not-inc', 'Private', 'Federal-gov', 'Local-gov', 'Self-emp-inc', 'Without-pay'])
-education_option = ['Preschool', '1st-4th', '5th-6th', '7th-8th', '9th', '10th', '11th', '12th', 'HS-grad', 'Some-college', 'Assoc-voc', 'Assoc-acdm', 'Bachelors', 'Masters', 'Prof-school', 'Doctorate']
-marital_status_option = sorted(['Never-married', 'Married-civ-spouse', 'Divorced', 'Separated', 'Married-AF-spouse', 'Widowed', 'Married-spouse-absent'])
-occupation_option = sorted(['Adm-clerical', 'Exec-managerial', 'Handlers-cleaners','Prof-specialty', 'Sales', 'Farming-fishing', 'Machine-op-inspct','Other-service', 'Transport-moving', 'Tech-support','Craft-repair', 'Protective-serv', 'Armed-Forces','Priv-house-serv'])
-relationship_option = sorted(['Not-in-family', 'Husband', 'Wife', 'Own-child', 'Unmarried', 'Other-relative'])
-race_option = sorted(['White', 'Black', 'Other', 'Asian-Pac-Islander', 'Amer-Indian-Eskimo'])
+
+[("SVM - Jerome Agius", 0), ("Logistic Regression - Isaac Muscat", 1), ("Random Forest - Kyle Demicoli", 2)]
+
+workclass_options = [('State Government', 'State-gov'), 
+                    ('Self Employed Not Incorporated', 'Self-emp-not-inc'), 
+                    'Private', ('Federal Government', 'Federal-gov'), ('Local Government', 'Local-gov'), ('Self Employed Incorporated', 'Self-emp-inc'), ('Without Pay', 'Without-pay')]
+
+education_option = [('Pre-School', 'Preschool'), '1st-4th', '5th-6th', '7th-8th', '9th', '10th', '11th', '12th', ('High School Graduate', 'HS-grad'), ('Collage', 'Some-college'), ('Associate Degree - Vocational', 'Assoc-voc'), ('Associate Degree - Academic', 'Assoc-acdm'), 'Bachelors', 'Masters', ('Professional School', 'Prof-school'), 'Doctorate']
+
+marital_status_option = [('Never Married','Never-married'), ('Married Civilian Spouse', 'Married-civ-spouse'), 'Divorced', 'Separated', ('Married Armed Forces Spouse', 'Married-AF-spouse'), 'Widowed', ('Married Spouse Absent', 'Married-spouse-absent')]
+occupation_option = [('Administrative Clerical', 'Adm-clerical'), ('Executive Managerial', 'Exec-managerial'), ('Handlers and Cleaners', 'Handlers-cleaners'), ('Professional Specialty', 'Prof-specialty'), 'Sales', ('Farming and Fishing', 'Farming-fishing'), ('Machine Operator and Inspector', 'Machine-op-inspct'), ('Other Service', 'Other-service'), ('Transport and Moving', 'Transport-moving'), ('Technical Support', 'Tech-support'), ('Craft and Repair', 'Craft-repair'), ('Protective Services', 'Protective-serv'), ('Armed Forces', 'Armed-Forces'), ('Private Household Services' ,'Priv-house-serv')]
+relationship_option = [('Not In Family', 'Not-in-family'), 'Husband', 'Wife', ('Biological Child', 'Own-child'), 'Unmarried', ('Other Relative', 'Other-relative')]
+race_option = ['White', 'Black', 'Other', ('Asian', 'Asian-Pac-Islander'), ('Indian', 'Amer-Indian-Eskimo')]
 sex_option = sorted(['Male', 'Female'])
 age = [0, 100]
 capital_gain = [0, 99999]
@@ -126,9 +134,43 @@ def Salary(model, workclass, education, marital_status, occupation, relationship
     # Make predictions with the loaded model
     prediction = loaded_model.predict(formattedDF)
 
+    probability = loaded_model.predict_proba(formattedDF)
+
+    # Get the number of classes
+    num_classes = probability.shape[1]
+
+    class_dict = {
+        0: '<=50K',
+        1: '>50K'
+    }
+
+    # Select the probabilities for a single sample (e.g., the first sample)
+    probabilities = probability[0] 
+
+    class_labels = [class_dict[i] for i in range(num_classes)]
+    colors = plt.cm.viridis(np.linspace(0, 1, num_classes))  # Use a colormap for consistent colors
+
+    fig, ax = plt.subplots(figsize=(10, 10))
+    _, _, autotexts = ax.pie(probabilities, colors=colors, autopct='%1.1f%%', startangle=140, pctdistance=1.1)
+
+    # Create a legend with colored boxes
+    legend_elements = []
+    for i, (color, label) in enumerate(zip(colors, class_labels)):
+        legend_elements.append(plt.Rectangle((0, 0), 1, 1, color=color, label=label))
+
+    ax.legend(handles=legend_elements, loc='upper left')
+    ax.set_title("Predicted Class Probabilities")
+
+    for i, p in enumerate(probabilities):
+      prob = float(round(p*100, 2))
+      if prob > 0:
+          autotexts[i].set_text(f"{prob}%")
+      else:
+          autotexts[i].set_text('')
+
     salary_result = '<=50K' if prediction[0] == 0 else '>50K'
 
-    return f"Predicted using {model_used} Salary Class: {salary_result}"
+    return f"Predicted using {model_used} Salary Class: {salary_result}", fig
 
 def Health(model, age, sex, bmi, children, smoker, region):
 
@@ -199,7 +241,44 @@ def Health(model, age, sex, bmi, children, smoker, region):
     prediction = loaded_model.predict(formattedDF)[0]
     prediction = inverse_mapping_charges[prediction]
 
-    return f"Predicted using {model_used} Charges Class: {prediction}"
+    probability = loaded_model.predict_proba(formattedDF)   
+    
+    # Get the number of classes
+    num_classes = probability.shape[1]
+
+    class_dict = {
+        0: 'Very Low (<= 5000)',
+        1: 'Low (5001 - 10000)',
+        2: 'Moderate (10001 - 15000)',
+        3: 'High (15001 - 20000)',
+        4: 'Very High (> 20001)',
+    }
+
+    # Select the probabilities for a single sample (e.g., the first sample)
+    probabilities = probability[0] 
+
+    class_labels = [class_dict[i] for i in range(num_classes)]
+    colors = plt.cm.viridis(np.linspace(0, 1, num_classes))  # Use a colormap for consistent colors
+
+    fig, ax = plt.subplots(figsize=(10, 10))
+    _, _, autotexts = ax.pie(probabilities, colors=colors, autopct='%1.1f%%', startangle=140, pctdistance=1.1)
+
+    # Create a legend with colored boxes
+    legend_elements = []
+    for i, (color, label) in enumerate(zip(colors, class_labels)):
+        legend_elements.append(plt.Rectangle((0, 0), 1, 1, color=color, label=label))
+
+    ax.legend(handles=legend_elements, loc='upper left')
+    ax.set_title("Predicted Class Probabilities")
+
+    for i, p in enumerate(probabilities):
+      prob = float(round(p*100, 2))
+      if prob > 0:
+          autotexts[i].set_text(f"{prob}%")
+      else:
+          autotexts[i].set_text('')
+
+    return f"Predicted using {model_used} Charges Class: {prediction}", fig
 
 # interface one
 iface1 = gr.Interface(
@@ -218,8 +297,9 @@ iface1 = gr.Interface(
         gr.Slider(minimum=capital_loss[0], maximum=capital_loss[1], step=1, label="Capital Loss"),
         gr.Slider(minimum=hours_per_week[0], maximum=hours_per_week[1], step=1, label="Hours per Week"),
     ],
-    outputs="text",
-    title="SVM - Salary"
+    outputs=[gr.Text(label="Predicted Label"), gr.Plot(label="Predicted Class Probabilities")],
+    title="SVM - Salary",
+    flagging_mode="never"
 )
 
 # interface two
@@ -234,8 +314,9 @@ iface2 = gr.Interface(
         gr.Dropdown(choices=smoker_option, label="Smoker"),
         gr.Dropdown(choices=region_option, label="Region"),
     ],
-    outputs="text",
-    title="SVM - Health"
+    outputs=[gr.Text(label="Predicted Label"), gr.Plot(label="Predicted Class Probabilities")],
+    title="SVM - Health",
+    flagging_mode="never"
 )
 
 demo = gr.TabbedInterface([iface1, iface2], ["Salary Prediction", "Health Charges Prediction"])