app.py

import streamlit as st
import pandas as pd
import requests
import plotly.express as px
import matplotlib.font_manager as fm
import matplotlib as mpl
import io
import time
from sklearn.cluster import KMeans
from sklearn.preprocessing import StandardScaler

# 確保正確的中文字符編碼
st.set_page_config(page_title="🌳台灣中小企業ESG數據分析與揭露儀表板🌲", page_icon=":chart_with_upwards_trend:", layout="wide")

# 定義 URL
urls = {
    "溫室氣體": "https://mopsfin.twse.com.tw/opendata/t187ap46_L_1.csv",
    "能源": "https://mopsfin.twse.com.tw/opendata/t187ap46_O_2.csv",
    "董事會揭露": "https://mopsfin.twse.com.tw/opendata/t187ap46_L_6.csv"
}

# 下載並載入 CSV 檔案到 DataFrame 的函數
@st.cache_data
def load_data(url):
    response = requests.get(url)
    response.encoding = 'utf-8'
    df = pd.read_csv(io.StringIO(response.text), encoding='utf-8')
    df = df.fillna(0)
    return df

# Streamlit 應用程式
st.title("台灣企業ESG數據分析與揭露")
st.subheader("以溫室氣體 X 再生能源 X 董事會資訊: https://www.tejwin.com/insight/carbon-footprint-verification/")
st.subheader("ESG投資: https://www.fhtrust.com.tw/ESG/operating")

# 允許用戶選擇數據集
dataset_choice = st.selectbox("選擇要顯示的數據集", list(urls.keys()))

# 載入選定的數據集
selected_df = load_data(urls[dataset_choice])

# 顯示爬取的資料
st.write("### 爬取的資料預覽")
st.dataframe(selected_df.head())

# 過濾出數值類型的欄位，排除 '出表日期' 和 '報告年度'
numeric_columns = selected_df.select_dtypes(include=['float64', 'int64']).columns
numeric_columns = [col for col in numeric_columns if col not in ['出表日期', '報告年度']]

# 允許用戶選擇用於繪製圖表的欄位
column_choice = st.selectbox("選擇欄位來繪製圖表", numeric_columns)

# 添加一個生成圖表的按鈕
if st.button("生成圖表"):
    # 顯示進度條
    progress_bar = st.progress(0)
    for i in range(100):
        time.sleep(0.01)
        progress_bar.progress(i + 1)

    # 創建一個標籤頁佈局
    tab1, tab2, tab3 = st.tabs(["圓餅圖", "長條圖", "K-means分析"])

    with tab1:
        # 使用 plotly 創建圓餅圖
        fig_pie = px.pie(
            selected_df,
            names='公司名稱',
            values=column_choice,
            title=f"{dataset_choice} - {column_choice} 圓餅圖",
            color_discrete_sequence=px.colors.qualitative.Pastel
        )
        fig_pie.update_traces(textposition='inside', textinfo='percent+label')
        fig_pie.update_layout(
            font=dict(size=12),
            legend=dict(
                orientation="h",
                yanchor="top",
                y=-0.3,
                xanchor="center",
                x=0.5
            ),
            height=700,
            margin=dict(t=50, b=50, l=50, r=50)
        )
        st.plotly_chart(fig_pie, use_container_width=True)

    with tab2:
        # 使用 plotly 創建長條圖
        fig_bar = px.bar(
            selected_df,
            x='公司名稱',
            y=column_choice,
            title=f"{dataset_choice} - {column_choice} 長條圖",
            color='公司名稱',
            color_discrete_sequence=px.colors.qualitative.Pastel
        )
        fig_bar.update_layout(
            xaxis_title="企業",
            yaxis_title=column_choice,
            font=dict(size=12),
            xaxis_tickangle=-45,
            showlegend=False,
            height=600
        )
        st.plotly_chart(fig_bar, use_container_width=True)

    with tab3:
        if dataset_choice == "溫室氣體":
            # 對溫室氣體數據進行K-means分析
            st.subheader("溫室氣體數據的K-means分析")

            # 選擇用於聚類的特徵
            cluster_features = st.multiselect("選擇用於聚類的特徵", numeric_columns, default=numeric_columns[:2])

            if len(cluster_features) >= 2:
                # 準備數據
                X = selected_df[cluster_features]
                scaler = StandardScaler()
                X_scaled = scaler.fit_transform(X)

                # 執行K-means聚類
                n_clusters = st.slider("選擇聚類數量", min_value=2, max_value=10, value=3)
                kmeans = KMeans(n_clusters=n_clusters, random_state=42)
                clusters = kmeans.fit_predict(X_scaled)

                # 添加聚類結果到數據框
                selected_df['Cluster'] = clusters

                # 視覺化聚類結果
                fig_scatter = px.scatter(
                    selected_df,
                    x=cluster_features[0],
                    y=cluster_features[1],
                    color='Cluster',
                    hover_data=['公司名稱'],
                    title=f"溫室氣體數據的K-means聚類 ({cluster_features[0]} vs {cluster_features[1]})"
                )
                st.plotly_chart(fig_scatter, use_container_width=True)

                # 顯示每個聚類的特徵
                st.subheader("聚類特徵")
                cluster_stats = selected_df.groupby('Cluster')[cluster_features].mean()
                st.dataframe(cluster_stats)

            else:
                st.warning("請至少選擇兩個特徵進行聚類分析。")
        else:
            st.info("K-means分析僅適用於溫室氣體數據集。")

    st.success("圖表生成完成！")

# 下載並設置自定義字體以顯示中文字符
font_url = "https://drive.google.com/uc?id=1eGAsTN1HBpJAkeVM57_C7ccp7hbgSz3_&export=download"
font_response = requests.get(font_url)
with open("TaipeiSansTCBeta-Regular.ttf", "wb") as font_file:
    font_file.write(font_response.content)
fm.fontManager.addfont("TaipeiSansTCBeta-Regular.ttf")
mpl.rc('font', family='Taipei Sans TC Beta')