app.py

import pandas as pd
import numpy as np
import joblib
import gradio as gr


# Load the preprocessing steps and the model
label_encoders = joblib.load("label_encoders.pkl")
one_hot_encoder = joblib.load("one_hot_encoder.pkl")
min_max_scaler = joblib.load("min_max_scaler.pkl")
model = joblib.load("logistic_regression_model.pkl")
le_target = joblib.load("label_encoder_target.pkl")


def preprocess_data(data):

    df = pd.DataFrame([data])

    label_encode_cols = [
        "Partner",
        "Dependents",
        "PhoneService",
        "PaperlessBilling",
        "gender",
    ]
    one_hot_encode_cols = [
        "MultipleLines",
        "InternetService",
        "OnlineSecurity",
        "OnlineBackup",
        "DeviceProtection",
        "TechSupport",
        "StreamingTV",
        "StreamingMovies",
        "Contract",
        "PaymentMethod",
    ]
    min_max_scale_cols = ["tenure", "MonthlyCharges", "TotalCharges"]

    # Strip leading and trailing spaces from string inputs
    for col in label_encode_cols + one_hot_encode_cols:
        df[col] = df[col].str.strip()

    # Convert non-numeric values to NaN and fill them with the mean of the column
    df[min_max_scale_cols] = df[min_max_scale_cols].replace(" ", np.nan).astype(float)

    df[min_max_scale_cols] = df[min_max_scale_cols].fillna(
        df[min_max_scale_cols].mean()
    )

    # Label encode specified columns
    for col in label_encode_cols:
        le = label_encoders[col]
        df[col] = le.transform(df[col])

    # One-hot encode specified columns
    one_hot_encoded = one_hot_encoder.transform(df[one_hot_encode_cols])

    # Min-max scale specified columns
    scaled_numerical = min_max_scaler.transform(df[min_max_scale_cols])

    # Combine processed columns into one DataFrame
    X_processed = np.hstack(
        (df[label_encode_cols].values, scaled_numerical, one_hot_encoded)
    )

    return X_processed


def predict(
    gender,
    senior_citizen,
    partner,
    dependents,
    tenure,
    phone_service,
    multiple_lines,
    internet_service,
    online_security,
    online_backup,
    device_protection,
    tech_support,
    streaming_tv,
    streaming_movies,
    contract,
    paperless_billing,
    payment_method,
    monthly_charges,
    total_charges,
):

    data = {
        "gender": gender,
        "SeniorCitizen": senior_citizen,
        "Partner": partner,
        "Dependents": dependents,
        "tenure": tenure,
        "PhoneService": phone_service,
        "MultipleLines": multiple_lines,
        "InternetService": internet_service,
        "OnlineSecurity": online_security,
        "OnlineBackup": online_backup,
        "DeviceProtection": device_protection,
        "TechSupport": tech_support,
        "StreamingTV": streaming_tv,
        "StreamingMovies": streaming_movies,
        "Contract": contract,
        "PaperlessBilling": paperless_billing,
        "PaymentMethod": payment_method,
        "MonthlyCharges": monthly_charges,
        "TotalCharges": total_charges,
    }

    try:
        X_new = preprocess_data(data)
        prediction = model.predict(X_new)
        prediction = le_target.inverse_transform(prediction)
        return "Churn" if prediction[0] == "Yes" else "No Churn"
    except Exception as e:
        print("Error during prediction:", e)
        return str(e)


# Define the Gradio interface
inputs = [
    gr.Radio(label="Gender", choices=["Female", "Male"]),
    gr.Number(label="Senior Citizen (0 or 1)"),
    gr.Radio(label="Partner", choices=["Yes", "No"]),
    gr.Radio(label="Dependents", choices=["Yes", "No"]),
    gr.Number(label="Tenure (integer)"),
    gr.Radio(label="Phone Service", choices=["Yes", "No"]),
    gr.Radio(label="Multiple Lines", choices=["Yes", "No", "No phone service"]),
    gr.Radio(label="Internet Service", choices=["DSL", "Fiber optic", "No"]),
    gr.Radio(label="Online Security", choices=["Yes", "No", "No internet service"]),
    gr.Radio(label="Online Backup", choices=["Yes", "No", "No internet service"]),
    gr.Radio(label="Device Protection", choices=["Yes", "No", "No internet service"]),
    gr.Radio(label="Tech Support", choices=["Yes", "No", "No internet service"]),
    gr.Radio(label="Streaming TV", choices=["Yes", "No", "No internet service"]),
    gr.Radio(label="Streaming Movies", choices=["Yes", "No", "No internet service"]),
    gr.Radio(label="Contract", choices=["Month-to-month", "One year", "Two year"]),
    gr.Radio(label="Paperless Billing", choices=["Yes", "No"]),
    gr.Radio(
        label="Payment Method",
        choices=[
            "Electronic check",
            "Mailed check",
            "Bank transfer (automatic)",
            "Credit card (automatic)",
        ],
    ),
    gr.Number(label="Monthly Charges (float)"),
    gr.Number(label="Total Charges (float)"),
]

outputs = gr.Textbox(label="Prediction")

# Create the Gradio interface
gr.Interface(
    fn=predict, inputs=inputs, outputs=outputs, title="Churn Prediction Model"
).launch(share=True)