pages/10_Decision_Tree.py

import streamlit as st
from sklearn import tree
from sklearn.datasets import load_iris
from sklearn.tree import export_graphviz
import graphviz
import pandas as pd

st.set_page_config(page_title="Decision Tree Explorer", page_icon="🌳", layout="wide")

st.title("🌳 Decision Tree Algorithm Explorer")
st.write("Understand how Decision Trees work with simple explanations, visuals, and real-world examples.")

section = st.radio("Choose a topic to explore:", [
    "What is a Decision Tree?",
    "How It Works",
    "Entropy vs Gini",
    "Tree Construction",
    "Classification vs Regression",
    "Pruning",
    "Feature Importance",
    "Visualize Example Tree",
    "Try with Iris Data"
])


if section == "What is a Decision Tree?":
    st.header("📘 What is a Decision Tree?")
    st.markdown("""
    A **Decision Tree** is a flowchart-like model that makes decisions based on a series of questions.
    
    - 🎯 Used in both **classification** (e.g., spam vs. not spam) and **regression** (e.g., predicting price).
    - 🌱 It starts at a **root**, asks a question, and branches out based on answers.
    - 🔚 Ends at a **leaf node** — which is the prediction.
    
    **Real-life example:**  
    You're deciding what to wear. You ask:
    1. Is it raining?  
    2. Is it cold?  
    → Based on your answers, you decide: jacket, umbrella, or just a T-shirt.
    """)

elif section == "How It Works":
    st.header("⚙️ How Does It Work?")
    st.markdown("""
    **Step-by-step:**
    1. Start with the whole dataset.
    2. Choose the feature that best splits the data.
    3. Split the dataset.
    4. Repeat until you reach a stopping condition.
    
    **Used concepts:**
    - Entropy (information gain)
    - Gini impurity
    """)

elif section == "Entropy vs Gini":
    st.header("📊 Entropy vs Gini")
    st.markdown("""
    ### Entropy
    Measures randomness or disorder in data.
    $$
    H(Y) = - \sum p_i \log_2 p_i
    $$

    ### Gini Impurity
    Measures the probability of wrong classification.
    $$
    Gini(Y) = 1 - \sum p_i^2
    $$

    **Which to use?**  
    - Gini is faster → default in scikit-learn.  
    - Entropy gives more information-theoretic understanding.
    """)

elif section == "Tree Construction":
    st.header("🔧 How is the Tree Built?")
    st.markdown("""
    The tree is built **top-down** using a greedy algorithm:
    
    - Best feature is chosen using Gini or Entropy.
    - Splits continue until stopping criteria (e.g., max depth, pure leaf).
    
    **Tip**: Too many splits = overfitting!
    """)

elif section == "Classification vs Regression":
    st.header("📈 Classification vs Regression")
    st.markdown("""
    - **Classification Tree**: Predicts categories (Yes/No, Spam/Ham).
    - **Regression Tree**: Predicts continuous values (e.g., house price).
    
    **Example:**
    - Classification: Will a customer churn?  
    - Regression: What will be the next month’s sales?
    """)

elif section == "Pruning":
    st.header("✂️ Pruning Techniques")
    st.markdown("""
    **Why prune?**  
    To avoid overfitting by cutting unnecessary branches.
    
    ### Pre-Pruning
    - `max_depth`: limit depth
    - `min_samples_split`: split only if enough samples
    - `min_samples_leaf`: limit how small leaves can be

    ### Post-Pruning
    - Cost Complexity Pruning (using α)
    """)

elif section == "Feature Importance":
    st.header("📌 Feature Importance")
    st.markdown("""
    Decision Trees calculate how important each feature is by how much it reduces impurity.

    **Formula:**
    $$
    Importance = \frac{Total\ Gain\ from\ Feature}{Total\ Gain\ from\ All\ Features}
    $$

    Useful for feature selection and explaining model decisions.
    """)

elif section == "Visualize Example Tree":
    st.header("🌿 Visualize a Small Tree Example")
    iris = load_iris()
    clf = tree.DecisionTreeClassifier(max_depth=3)
    clf = clf.fit(iris.data, iris.target)
    dot_data = tree.export_graphviz(clf, out_file=None, 
                                feature_names=iris.feature_names,
                                class_names=iris.target_names,
                                filled=True, rounded=True,
                                special_characters=True)
    st.graphviz_chart(dot_data)


elif section == "Try with Iris Data":
    st.header("🌸 Try with Iris Dataset")
    iris = load_iris()
    df = pd.DataFrame(iris.data, columns=iris.feature_names)
    df['target'] = iris.target
    st.write("Here's a preview of the dataset:")
    st.dataframe(df.head())

    st.markdown("### Build and visualize a Decision Tree")
    max_depth = st.slider("Select max depth of the tree:", 1, 5, 3)
    clf = tree.DecisionTreeClassifier(max_depth=max_depth)
    clf = clf.fit(iris.data, iris.target)
    dot_data = tree.export_graphviz(clf, out_file=None,
                                    feature_names=iris.feature_names,
                                    class_names=iris.target_names,
                                    filled=True, rounded=True)
    st.graphviz_chart(dot_data)

st.markdown("---")
st.success("✅ Decision Trees are simple yet powerful! Tune them well, visualize their structure, and understand every split.")