Harshitha-01/LogisticRegression

1import streamlit as st2import pandas as pd3import numpy as np4import matplotlib.pyplot as plt5import seaborn as sns6from sklearn.datasets import load_iris7from sklearn.model_selection import train_test_split8from sklearn.linear_model import LogisticRegression9from sklearn.preprocessing import StandardScaler10from sklearn.metrics import accuracy_score, classification_report11 12# Streamlit Page Setup13st.set_page_config(page_title="Logistic Regression", layout="wide")14st.title("📈 Logistic Regression")15 16# Tabs17tab1, tab2, tab3 = st.tabs(["📘 Introduction", "🔬 Train Model", "📊 Results & Visualization"])18 19# ----------- TAB 1: INTRODUCTION -----------20with tab1:21    st.header("🤖 What is Logistic Regression?")22    st.markdown("""23    Logistic Regression is a **classification algorithm** that models the probability of class membership.24 25    ---26    ### 🧠 Key Concepts:27    - Computes a **weighted sum** of inputs28    - Applies the **sigmoid function** to produce probabilities29    - Uses a **threshold** (e.g. 0.5) for classification30    - Handles **multiclass problems** using One-vs-Rest or softmax31 32    ---33    ### ⚙ Advantages:34    - Simple and easy to implement35    - Interpretable outputs (probabilities)36    - Fast training, especially for linearly separable data37 38    ---39    ### 💡 When to Use?40    - Linearly separable data41    - Need probability estimates42    - As a baseline for other classifiers43    """)44 45# ----------- TAB 2: MODEL TRAINING -----------46with tab2:47    st.header("🌼 Try Logistic Regression on the Iris Dataset")48 49    # Load dataset50    iris = load_iris()51    df = pd.DataFrame(iris.data, columns=iris.feature_names)52    df["target"] = iris.target53    st.subheader("📄 Dataset Preview")54    st.dataframe(df.head(), use_container_width=True)55 56    # Model Parameters57    st.subheader("⚙️ Choose Model Parameters")58    col1, col2 = st.columns(2)59    with col1:60        penalty = st.radio("Penalty Type", ["l2", "none"])61    with col2:62        C = st.slider("Inverse Regularization (C)", 0.01, 10.0, value=1.0)63 64    # Preprocessing65    X = df.drop("target", axis=1)66    y = df["target"]67    scaler = StandardScaler()68    X_scaled = scaler.fit_transform(X)69 70    # Train/Test Split71    X_train, X_test, y_train, y_test = train_test_split(X_scaled, y, test_size=0.2, random_state=42)72 73    # Model Training74    model = LogisticRegression(penalty=penalty, C=C, solver='lbfgs', multi_class='ovr', max_iter=200)75    model.fit(X_train, y_train)76    y_pred = model.predict(X_test)77    acc = accuracy_score(y_test, y_pred)78 79    # Store for next tab80    st.session_state["model"] = model81    st.session_state["df"] = df82    st.session_state["scaler"] = scaler83    st.session_state["X"] = X84    st.session_state["y"] = y85    st.session_state["y_pred"] = y_pred86    st.session_state["y_test"] = y_test87    st.session_state["target_names"] = iris.target_names88 89    st.success(f"✅ Logistic Regression Model Accuracy: **{acc*100:.2f}%**")90 91# ----------- TAB 3: RESULTS & VISUALIZATION -----------92with tab3:93    st.header("📊 Results and Visualization")94 95    if "model" in st.session_state:96        st.subheader("📋 Classification Report")97        st.text(classification_report(98            st.session_state["y_test"],99            st.session_state["y_pred"],100            target_names=st.session_state["target_names"]101        ))102 103        st.subheader("🌌 Decision Boundary Visualization")104 105        # Feature selection106        df = st.session_state["df"]107        feature_x = st.selectbox("X-axis Feature", df.columns[:-1], index=0)108        feature_y = st.selectbox("Y-axis Feature", df.columns[:-1], index=1)109 110        # Prepare Data111        X_vis = df[[feature_x, feature_y]]112        X_vis_scaled = st.session_state["scaler"].fit_transform(X_vis)113        X_train_v, X_test_v, y_train_v, y_test_v = train_test_split(X_vis_scaled, st.session_state["y"], test_size=0.2, random_state=42)114 115        model_vis = LogisticRegression(C=C, multi_class='ovr', solver='lbfgs', max_iter=200)116        model_vis.fit(X_train_v, y_train_v)117 118        # Mesh grid119        h = 0.02120        x_min, x_max = X_vis_scaled[:, 0].min() - 1, X_vis_scaled[:, 0].max() + 1121        y_min, y_max = X_vis_scaled[:, 1].min() - 1, X_vis_scaled[:, 1].max() + 1122        xx, yy = np.meshgrid(np.arange(x_min, x_max, h), np.arange(y_min, y_max, h))123        Z = model_vis.predict(np.c_[xx.ravel(), yy.ravel()])124        Z = Z.reshape(xx.shape)125 126        # Plotting127        fig, ax = plt.subplots(figsize=(5, 3))128        plt.contourf(xx, yy, Z, alpha=0.3, cmap='coolwarm')129        sns.scatterplot(x=X_vis_scaled[:, 0], y=X_vis_scaled[:, 1], hue=df["target"], palette="deep", ax=ax)130        plt.xlabel(feature_x)131        plt.ylabel(feature_y)132        plt.title("Logistic Regression Decision Boundaries")133        st.pyplot(fig)134    else:135        st.warning("⚠️ Please train the model in the 'Train Model' tab first.")136# Footer137st.markdown("---")138 139