app.py

import streamlit as st
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_squared_error, r2_score
from sklearn import datasets
import io

def main():
    st.title("California Housing Analysis")

    # Load the California housing dataset
    california = datasets.fetch_california_housing()
    df = pd.DataFrame(california.data, columns=california.feature_names)
    df['MedHouseVal'] = california.target

    # Displaying initial data information
    st.write("## Data Sample")
    st.write(df.head())

    st.write("## Data Statistics")
    st.write(df.describe())

    st.write("## Data Info")
    buffer = io.StringIO()
    df.info(buf=buffer)
    s = buffer.getvalue()
    st.text(s)

    st.write("## Missing Values")
    st.write(df.isnull().sum())

    # Fixed target variable
    target = 'MedHouseVal'
    st.write(f"## Target Variable: {target}")

    # Drop the target from the predictors list
    predictor_options = df.columns.drop(target).tolist()
    
    # Multiselect widget to select predictor variables for regression
    predictors = st.multiselect(
        'Select predictor variables for regression:',
        options=predictor_options,
        default=predictor_options  # default to all predictors for MLR
    )

    # Splitting data for regression
    X = df[predictors]
    y = df[target]
    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

    # Perform multilinear regression
    mlr_model = LinearRegression()
    mlr_model.fit(X_train, y_train)
    mlr_y_pred = mlr_model.predict(X_test)
    mlr_rmse = np.sqrt(mean_squared_error(y_test, mlr_y_pred))
    mlr_r2 = r2_score(y_test, mlr_y_pred)

    # Perform simple linear regression using only one predictor if possible
    if 'AveRooms' in predictors:
        slr_model = LinearRegression()
        slr_X_train = X_train[['AveRooms']]
        slr_X_test = X_test[['AveRooms']]
        slr_model.fit(slr_X_train, y_train)
        slr_y_pred = slr_model.predict(slr_X_test)
        slr_rmse = np.sqrt(mean_squared_error(y_test, slr_y_pred))
        slr_r2 = r2_score(y_test, slr_y_pred)

        # Display RMSE and R-squared comparisons
        st.write("## Regression Performance Comparison")
        st.write(f"### Multilinear Regression (using all selected predictors)")
        st.write(f'RMSE: {mlr_rmse}')
        st.write(f'R-squared: {mlr_r2}')
        
        st.write(f"### Simple Linear Regression (using 'AveRooms')")
        st.write(f'RMSE: {slr_rmse}')
        st.write(f'R-squared: {slr_r2}')

        # Plotting both regressions
        fig, ax = plt.subplots(1, 2, figsize=(15, 6))
        
        ax[0].scatter(y_test, mlr_y_pred, color='blue')
        ax[0].plot(y_test, y_test, color='red')
        ax[0].set_title('Multilinear Regression: Actual vs Predicted')
        ax[0].set_xlabel('Actual Values')
        ax[0].set_ylabel('Predicted Values')
        
        ax[1].scatter(y_test, slr_y_pred, color='green')
        ax[1].plot(y_test, y_test, color='red')
        ax[1].set_title("Simple Linear Regression ('AveRooms'): Actual vs Predicted")
        ax[1].set_xlabel('Actual Values')
        ax[1].set_ylabel('Predicted Values')
        
        st.pyplot(fig)

if __name__ == "__main__":
    main()