app.py

import streamlit as st
import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
from scipy.cluster.hierarchy import linkage, dendrogram, fcluster
from sklearn.preprocessing import StandardScaler

# Load dataset
@st.cache_data
def load_data():
    data = pd.read_csv("kpopidolsv3.csv")
    return data

data = load_data()

# Preprocess data
def preprocess_data(data):
    features = ['Height', 'Weight']
    df = data.dropna(subset=features)
    scaler = StandardScaler()
    scaled_features = scaler.fit_transform(df[features])
    return scaled_features, df

# Perform hierarchical clustering
def apply_hierarchical_clustering(scaled_features, method='ward'):
    Z = linkage(scaled_features, method=method)
    return Z

# Sidebar controls
st.sidebar.header("Clustering Parameters")
num_clusters = st.sidebar.slider("Number of Clusters", 2, 10, 3)

def main():
    st.title("🎤 K-Pop Idol Clustering using Hierarchical Clustering")

    # Tabs for Navigation
    tab1, tab2, tab3 = st.tabs(["📚 About the App", "📊 Dataset & Results", "🔎 Explore Idols"])

    with tab1:
        st.header("📚 About the App")
        st.markdown(
            "This app groups K-pop idols based on their physical features (height, weight), company, and debut information using **Hierarchical Clustering with Ward's Method**."
        )
        st.markdown(
            "### How It Works:
            - **Dendrogram Visualization:** Explore hierarchical clusters.
            - **Dynamic Cluster Cutting:** Set the number of clusters dynamically.
            - **Idol Comparison:** Analyze clusters by different features."
        )

    with tab2:
        st.header("📊 Dataset Overview and Results")
        st.write("### Sample Data")
        st.dataframe(data.head())

        # Preprocess and cluster
        scaled_features, df_processed = preprocess_data(data)
        Z = apply_hierarchical_clustering(scaled_features)

        # Dendrogram
        st.write("### Dendrogram")
        plt.figure(figsize=(12, 6))
        dendrogram(Z, labels=df_processed['Stage Name'].values, leaf_rotation=90)
        st.pyplot(plt)

        # Cut the dendrogram
        cluster_labels = fcluster(Z, num_clusters, criterion='maxclust')
        df_processed['Cluster'] = cluster_labels
        st.write("### Clustered Data Sample")
        st.dataframe(df_processed[['Stage Name', 'Company', 'Nationality', 'Cluster']].head(10))

    with tab3:
        st.header("🔎 Explore Idols by Company or Nationality")
        option = st.selectbox("Filter idols by:", ["Company", "Nationality"])
        selected_value = st.text_input(f"Enter {option} name:")
        
        if selected_value:
            filtered_data = df_processed[df_processed[option].str.contains(selected_value, na=False, case=False)]
            if not filtered_data.empty:
                st.dataframe(filtered_data[['Stage Name', 'Company', 'Nationality', 'Cluster']])
            else:
                st.warning(f"No idols found for {option}: {selected_value}")

if __name__ == "__main__":
    main()