app.py

import pandas as pd
import matplotlib.pyplot as plt
%matplotlib inline
churn=pd.read_csv(r"C:\Users\deepu\Downloads\archive (9)\churn-bigml-80.csv")
churn
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt


categorical_cols = churn.select_dtypes(include=['object']).columns.tolist()

# One-hot encode categorical columns
data_encoded = pd.get_dummies(churn, columns=categorical_cols, drop_first=True)

# Calculate correlations with the target variable (assuming the target column is named 'churn')
target = 'Churn'
correlations = data_encoded.corr()[target].drop(target)  # Drop target's self-correlation

# Sort correlations by absolute value (strongest to weakest correlation)
correlations = correlations.sort_values(key=abs, ascending=False)

# Plot the correlations as a bar chart
plt.figure(figsize=(10, 8))
sns.barplot(x=correlations.values, y=correlations.index, palette="coolwarm")
plt.title("Feature Correlations with Churn")
plt.xlabel("Correlation Coefficient")
plt.ylabel("Features")
plt.show()
import pandas as pd


# Identify categorical columns
categorical_cols = churn.select_dtypes(include=['object']).columns.tolist()

# One-hot encode categorical columns
data_encoded = pd.get_dummies(churn, columns=categorical_cols, drop_first=True)

# Calculate correlations with the target variable (assuming the target column is named 'churn')
target = 'Churn'
correlations = data_encoded.corr()[target].drop(target)  # Drop self-correlation of target

# Filter features with correlation > 0.1 or < -0.05
filtered_features = correlations[(correlations > 0.1) | (correlations < -0.05)].index.tolist()

# Create a new DataFrame with only the filtered features and the target column
data_filtered = data_encoded[filtered_features + [target]]

# Save the filtered DataFrame to a CSV file
data_filtered.to_csv("filtered_features_churn.csv", index=False)

print("CSV file 'filtered_features_churn.csv' created successfully with filtered features.")
data=pd.read_csv(r"filtered_features_churn.csv")
data = data.drop(columns=["Churn"])
import gradio as gr
import numpy as np
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split

X = data.drop(columns=["Churn"])
y =data["Churn"]
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
model = RandomForestClassifier(random_state=42)
model.fit(X_train, y_train)
import gradio as gr
import numpy as np
from sklearn.preprocessing import StandardScaler
from sklearn.ensemble import RandomForestClassifier


# Initialize and fit the scaler
scaler = StandardScaler()
scaler.fit(X_train)  # Fit the scaler to your training data

# Initialize and fit your model
model = RandomForestClassifier()
model.fit(X_train, y_train)  # Fit the model to your training data

# Define a prediction function
def predict_churn(*features):
    input_data = np.array(features).reshape(1, -1)
    input_data_scaled = scaler.transform(input_data)  # Use the fitted scaler here
    
    # Predict churn probability
    prediction = model.predict(input_data_scaled)
    return "Churn" if prediction[0] == 1 else "Not Churn"

# Example feature labels (adjust based on your dataset)
feature_labels = ['Number vmail messages', 'Total day minutes', 'Total day charge',
       'Total intl calls', 'Customer service calls', 'International plan_Yes',
       'Voice mail plan_Yes']  # Replace with your actual feature names

# Set up Gradio interface
interface = gr.Interface(
    fn=predict_churn,
    inputs=[gr.Number(label=label) for label in feature_labels],
    outputs="text",
    title="Customer Churn Prediction",
    description="Enter customer information to predict churn .",
    theme="soft",
    flagging_options=["average prediction", "good prediction", "bad prediction"]

)

# Launch the interface
interface.launch(share=True)