import streamlit as st import pandas as pd import numpy as np import matplotlib.pyplot as plt import plotly.express as px from wordcloud import WordCloud import os import math def render_visualisasi(): st.title("📈 Dashboard Visualisasi Data") st.markdown("Analisis visual interaktif terhadap data opini publik terkait kebijakan anggaran pendidikan.") # ============================================================================== # 1. LOAD DATA UTAMA # ============================================================================== file_path = 'data/Data_Lengkap_Tokenisasi.csv' if not os.path.exists(file_path): st.error(f"❌ File dataset tidak ditemukan di: {file_path}") return # Load Data df = pd.read_csv(file_path) if 'Label' in df.columns: df['Label_Clean'] = df['Label'].astype(str).str.lower().str.strip() else: st.error("❌ Kolom 'Label' tidak ditemukan dalam CSV.") return if 'created_at' in df.columns: df['Tanggal'] = pd.to_datetime(df['created_at']).dt.date elif 'Tanggal' in df.columns: df['Tanggal'] = pd.to_datetime(df['Tanggal']).dt.date else: st.warning("⚠ī¸ Kolom tanggal tidak ditemukan. Grafik tren waktu mungkin tidak muncul.") # ============================================================================== # 2. VISUALISASI DISTRIBUSI SENTIMEN (PIE & BAR) # ============================================================================== st.subheader("📊 Distribusi & Polaritas Sentimen") col_pie, col_bar = st.columns([1, 1.5]) # --- A. PIE CHART --- with col_pie: df_pie = df['Label_Clean'].value_counts().reset_index() df_pie.columns = ['Sentimen', 'Jumlah'] fig_pie = px.pie( df_pie, names='Sentimen', values='Jumlah', hole=0.4, color='Sentimen', color_discrete_map={'negatif':'#FF4B4B', 'netral':'#808495', 'positif':'#00CC96'}, title="Persentase Sentimen" ) fig_pie.update_layout(showlegend=True, legend=dict(orientation="h", yanchor="bottom", y=-0.3, xanchor="center", x=0.5)) st.plotly_chart(fig_pie, use_container_width=True) # --- B. TREN WAKTU --- with col_bar: if 'Tanggal' in df.columns: start_date = pd.to_datetime("2025-02-01").date() end_date = pd.to_datetime("2025-03-31").date() df_filtered = df[ (df['Tanggal'] >= start_date) & (df['Tanggal'] <= end_date) ] kolom_label = 'Label' if 'Label' in df_filtered.columns else 'Label_Clean' df_trend = df_filtered.groupby(['Tanggal', kolom_label]).size().reset_index(name='Jumlah') fig_trend = px.line( df_trend, x='Tanggal', y='Jumlah', color=kolom_label, markers=True, color_discrete_map={ 'negatif':'#FF4B4B', 'netral':'#808495', 'positif':'#00CC96', 'Negatif':'#FF4B4B', 'Netral':'#808495', 'Positif':'#00CC96', 'negative':'#FF4B4B', 'neutral':'#808495', 'positive':'#00CC96' }, title="Tren Sentimen Harian (Feb - Mar 2025)" ) fig_trend.update_xaxes(range=[start_date, end_date]) fig_trend.update_layout(xaxis_title="Tanggal", yaxis_title="Jumlah Tweet", hovermode="x unified", legend=dict(orientation="h", y=1.1)) st.plotly_chart(fig_trend, use_container_width=True) else: st.info("Data Tanggal tidak tersedia untuk menampilkan tren.") # ============================================================================== # 3. WORDCLOUD # ============================================================================== st.subheader("☁ī¸ WordCloud: Representasi Visual Teks") st.write("Kata-kata yang paling sering muncul dalam setiap kategori.") # 1. Fungsi Asli untuk generate dari Teks (Data Mentah & Bersih) def generate_wc(text, colormap): if not isinstance(text, str) or not text.strip(): st.warning("⚠ī¸ Tidak ada data teks yang cukup.") return with st.spinner("Sedang menggambar WordCloud..."): try: wc = WordCloud(width=800, height=400, background_color='white', colormap=colormap, max_words=100).generate(text) fig, ax = plt.subplots(figsize=(10, 5)) ax.imshow(wc, interpolation='bilinear') ax.axis("off") st.pyplot(fig) except Exception as e: st.error(f"Error WordCloud: {e}") # 2. FUNGSI BARU: Generate WordCloud langsung dari CSV Frekuensi agar instan def generate_wc_from_freq(file_path, colormap): if os.path.exists(file_path): try: df_freq = pd.read_csv(file_path) # Mengubah format DataFrame menjadi Dictionary (Syarat mutlak WordCloud) freq_dict = dict(zip(df_freq['Word'], df_freq['Frequency'])) with st.spinner("Merender WordCloud instan dari CSV..."): wc = WordCloud(width=800, height=400, background_color='white', colormap=colormap, max_words=100) wc.generate_from_frequencies(freq_dict) fig, ax = plt.subplots(figsize=(10, 5)) ax.imshow(wc, interpolation='bilinear') ax.axis("off") st.pyplot(fig) except Exception as e: st.error(f"Error memproses file CSV WordFreq: {e}") else: st.warning(f"⚠ī¸ File frekuensi belum tersedia: {file_path}") # Tabs Navigasi WordCloud tab_mentah, tab_bersih, tab_neg, tab_net, tab_pos = st.tabs([ "Data Mentah", "Data Bersih", "Negatif", "Netral", "Positif" ]) with tab_mentah: st.caption("Data dari kolom 'Teks Tweet' (Original)") generate_wc(" ".join(df['Teks Tweet'].dropna().astype(str)), 'cividis') with tab_bersih: st.caption("Data dari kolom 'Tweet_Final' (Preprocessed)") if 'Tweet_Final' in df.columns: generate_wc(" ".join(df['Tweet_Final'].dropna().astype(str)), 'viridis') else: st.warning("Kolom Tweet_Final tidak ada.") # MENGGUNAKAN FILE CSV WORDFREQ DI SINI with tab_neg: st.caption("Kata dominan sentimen NEGATIF (Sumber: WordFreq_Negatif.csv)") generate_wc_from_freq('model/WordFreq_Negatif.csv', 'Reds') with tab_net: st.caption("Kata dominan sentimen NETRAL (Sumber: WordFreq_Netral.csv)") generate_wc_from_freq('model/WordFreq_Netral.csv', 'Greys') with tab_pos: st.caption("Kata dominan sentimen POSITIF (Sumber: WordFreq_Positif.csv)") generate_wc_from_freq('model/WordFreq_Positif.csv', 'Greens') st.markdown("---") # ============================================================================== # 4. TOPIC MODELING # ============================================================================== st.subheader("📌 4. Topic Modeling (LDA) & Kata Kunci") st.write("Ekstraksi topik dominan dari hasil algoritma Latent Dirichlet Allocation (LDA).") path_lda = 'model/Hasil_Analisis_Topik_LDA.csv' if not os.path.exists(path_lda): path_lda = 'Hasil_Analisis_Topik_LDA.csv' if os.path.exists(path_lda): try: df_lda = pd.read_csv(path_lda) def parse_lda_string(text_data): data_items = [] for word in str(text_data).split(','): word = word.strip() if word: data_items.append({'Kata': word}) df_res = pd.DataFrame(data_items) if not df_res.empty: df_res['Bobot'] = range(len(df_res), 0, -1) df_res = df_res.sort_values(by='Bobot', ascending=True) return df_res t_neg, t_net, t_pos = st.tabs(["🔴 Topik Negatif", "âšĒ Topik Netral", "đŸŸĸ Topik Positif"]) mapping = {'negatif': t_neg, 'netral': t_net, 'positif': t_pos} for sentimen, tab in mapping.items(): with tab: df_subset = df_lda[df_lda['Sentimen'].str.lower() == sentimen] if df_subset.empty: st.warning(f"Belum ada data topik untuk {sentimen}.") else: for idx, row in df_subset.iterrows(): topik_ke = row['Topik Ke'] df_chart = parse_lda_string(row['Kata Kunci']) if not df_chart.empty: fig = px.bar( df_chart, x='Bobot', y='Kata', orientation='h', title=f"Topik {topik_ke}", color='Bobot', color_continuous_scale='Reds' if sentimen == 'negatif' else 'Greys' if sentimen == 'netral' else 'Greens' ) fig.update_layout(height=300, showlegend=False, xaxis_title=None, xaxis_visible=False) st.plotly_chart(fig, use_container_width=True) st.divider() except Exception as e: st.error(f"Gagal memproses data LDA: {e}") else: st.warning("⚠ī¸ File 'Hasil_Analisis_Topik_LDA.csv' belum tersedia di folder model.") # ============================================================================== # 5. DATA EXPLORER & EVALUASI MODEL # ============================================================================== st.subheader("🔍 Data Explorer & Evaluasi Model") tab_data, tab_eval = st.tabs(["Data Explorer", "Tabel Performa (Evaluasi)"]) # --- TAB 1: DATA EXPLORER --- with tab_data: col_f1, col_f2 = st.columns([1, 2]) with col_f1: filter_label = st.selectbox("Filter Sentimen:", ['Semua', 'negatif', 'netral', 'positif']) with col_f2: search_keyword = st.text_input("Cari Tweet:", "") cols_available = [c for c in ['created_at', 'username', 'Teks Tweet', 'Label_Clean'] if c in df.columns] df_show = df[cols_available].copy() rename_map = {'created_at': 'Tanggal', 'username': 'Username', 'Label_Clean': 'Label'} df_show = df_show.rename(columns=rename_map) if filter_label != 'Semua' and 'Label' in df_show.columns: df_show = df_show[df_show['Label'] == filter_label] if search_keyword and 'Teks Tweet' in df_show.columns: df_show = df_show[df_show['Teks Tweet'].str.contains(search_keyword, case=False, na=False)] df_show.index = range(1, len(df_show) + 1) baris_per_halaman = 20 total_data = len(df_show) total_halaman = math.ceil(total_data / baris_per_halaman) if total_data > 0: c_nav, c_stat = st.columns([1, 3]) with c_nav: halaman = st.number_input("Halaman", min_value=1, max_value=max(1, total_halaman), step=1) with c_stat: st.write("") st.caption(f"Menampilkan **{total_data}** Data (Halaman {halaman} dari {total_halaman})") start_idx = (halaman - 1) * baris_per_halaman end_idx = start_idx + baris_per_halaman df_page = df_show.iloc[start_idx:end_idx] st.dataframe(df_page, use_container_width=True) else: st.warning("Data tidak ditemukan.") # --- TAB 2: TABEL EVALUASI & CONFUSION MATRIX --- with tab_eval: st.subheader("1. Tabel Performa (Classification Report)") st.markdown(""" Metrik evaluasi model berdasarkan data testing (20%): * **Precision**: Ketepatan tebakan. * **Recall**: Kemampuan menemukan data yang relevan. * **F1-Score**: Rata-rata harmonis (Paling penting untuk data tidak seimbang). """) path_perf = 'model/Tabel_Performa_LSTM.csv' if not os.path.exists(path_perf): path_perf = 'Tabel_Performa_LSTM.csv' if os.path.exists(path_perf): try: df_perf = pd.read_csv(path_perf, index_col=0) st.table( df_perf.style.highlight_max(axis=0, props='background-color: #FFEB3B; color: black; font-weight: bold') ) except Exception as e: st.error(f"Gagal memuat tabel evaluasi: {e}") else: st.warning("⚠ī¸ File 'Tabel_Performa_LSTM.csv' belum tersedia.") st.divider() st.subheader("2. Confusion Matrix") st.markdown("Visualisasi ini menunjukkan **detail kesalahan prediksi**. Sumbu Y adalah Label Asli, Sumbu X adalah Prediksi Model.") path_cm = 'model/Data_Confusion_Matrix.csv' if not os.path.exists(path_cm): path_cm = 'Data_Confusion_Matrix.csv' if os.path.exists(path_cm): try: df_cm_data = pd.read_csv(path_cm) if 'y_true' in df_cm_data.columns and 'y_pred' in df_cm_data.columns: from sklearn.metrics import confusion_matrix labels = ['Negatif', 'Netral', 'Positif'] cm = confusion_matrix(df_cm_data['y_true'], df_cm_data['y_pred']) fig_cm = px.imshow( cm, text_auto=True, labels=dict(x="Prediksi Model", y="Label Aktual (Asli)", color="Jumlah Data"), x=labels, y=labels, color_continuous_scale='Blues', aspect="auto" ) fig_cm.update_layout(title="Confusion Matrix Heatmap") st.plotly_chart(fig_cm, use_container_width=True) total_benar = np.trace(cm) total_data = np.sum(cm) akurasi_cm = (total_benar / total_data) * 100 st.caption(f"💡 **Interpretasi:** Dari total **{total_data}** data testing, model berhasil menebak benar sebanyak **{total_benar}** data ({akurasi_cm:.2f}%).") else: st.error("Format CSV Confusion Matrix salah. Harus ada kolom 'y_true' dan 'y_pred'.") except Exception as e: st.error(f"Gagal memproses Confusion Matrix: {e}") else: st.info("ℹī¸ **Data Confusion Matrix belum tersedia.** Silakan jalankan kode penyimpanan `Data_Confusion_Matrix.csv` di Google Colab (Bagian Evaluasi).")