Spaces:
Sleeping
Sleeping
| import gradio as gr | |
| import pandas as pd | |
| import numpy as np | |
| from sklearn.preprocessing import LabelEncoder, StandardScaler | |
| from sklearn.neighbors import NearestNeighbors | |
| # Load data | |
| citizen_data = pd.read_csv("citizen_data_50000.csv") | |
| services_data = pd.read_csv("citizen_services_50000.csv") | |
| merged_data = pd.merge(citizen_data, services_data, on="citizen_id") | |
| # Encode categorical features including location | |
| categorical_cols = ['gender', 'education', 'marital_status', 'social_category', | |
| 'village', 'block', 'district'] | |
| encoders = {} | |
| for col in categorical_cols: | |
| le = LabelEncoder() | |
| merged_data[col] = le.fit_transform(merged_data[col]) | |
| encoders[col] = le | |
| # Normalize numeric columns | |
| scaler = StandardScaler() | |
| merged_data[['age', 'annual_income']] = scaler.fit_transform(merged_data[['age', 'annual_income']]) | |
| # Define feature columns including location | |
| feature_cols = ['age', 'gender', 'education', 'marital_status', 'social_category', | |
| 'annual_income', 'village', 'block', 'district'] | |
| features_matrix = merged_data[feature_cols].values | |
| # Train KNN model | |
| knn = NearestNeighbors(n_neighbors=10, metric='cosine') | |
| knn.fit(features_matrix) | |
| # Recommendation function with location | |
| def recommend(age, gender, education, marital, social, income, village, block, district): | |
| input_data = { | |
| 'age': [age], | |
| 'gender': [encoders['gender'].transform([gender])[0]], | |
| 'education': [encoders['education'].transform([education])[0]], | |
| 'marital_status': [encoders['marital_status'].transform([marital])[0]], | |
| 'social_category': [encoders['social_category'].transform([social])[0]], | |
| 'annual_income': [income], | |
| 'village': [encoders['village'].transform([village])[0]], | |
| 'block': [encoders['block'].transform([block])[0]], | |
| 'district': [encoders['district'].transform([district])[0]], | |
| } | |
| input_vector = pd.DataFrame(input_data) | |
| # Normalize numeric columns | |
| input_vector[['age', 'annual_income']] = scaler.transform(input_vector[['age', 'annual_income']]) | |
| # Get neighbors and calculate weighted similarity | |
| distances, indices = knn.kneighbors(input_vector.values, n_neighbors=10) | |
| similarities = 1 - distances[0] | |
| top_users = merged_data.iloc[indices[0]].copy() | |
| top_users['similarity_weight'] = similarities | |
| service_cols = [col for col in services_data.columns if col != "citizen_id"] | |
| for col in service_cols: | |
| top_users[col] *= top_users['similarity_weight'] | |
| recommended_services = top_users[service_cols].sum().sort_values(ascending=False).head(5) | |
| return [f"{service} ✅" for service in recommended_services.index] | |
| # Gradio UI with location inputs | |
| iface = gr.Interface( | |
| fn=recommend, | |
| inputs=[ | |
| gr.Slider(18, 100, value=30, label="Age"), | |
| gr.Dropdown(encoders['gender'].classes_.tolist(), label="Gender"), | |
| gr.Dropdown(encoders['education'].classes_.tolist(), label="Education Level"), | |
| gr.Dropdown(encoders['marital_status'].classes_.tolist(), label="Marital Status"), | |
| gr.Dropdown(encoders['social_category'].classes_.tolist(), label="Social Category"), | |
| gr.Slider(0, 2000000, value=300000, step=10000, label="Annual Income"), | |
| gr.Dropdown(encoders['village'].classes_.tolist(), label="Village"), | |
| gr.Dropdown(encoders['block'].classes_.tolist(), label="Block"), | |
| gr.Dropdown(encoders['district'].classes_.tolist(), label="District"), | |
| ], | |
| outputs=gr.List(label="Top 5 Recommended Services"), | |
| title="🧠 Citizen Service Recommender", | |
| description="Get recommended services based on citizens similar to you — using your demographic and location data.", | |
| ) | |
| iface.launch() |