Marcel0123/unsupervised-training
0
1 2import gradio as gr3import numpy as np4import pandas as pd5import matplotlib.pyplot as plt6from sklearn import datasets7from sklearn.preprocessing import StandardScaler8from sklearn.decomposition import PCA9import plotly.graph_objects as go10import plotly.express as px11import time12 13FEATURE_LABELS = {14 "age": "Leeftijd",15 "sex": "Geslacht",16 "bmi": "BMI (Body Mass Index)",17 "bp": "Bloeddruk",18 "s1": "Totale cholesterol",19 "s2": "LDL-cholesterol",20 "s3": "HDL-cholesterol",21 "s4": "Chol./HDL-verhouding",22 "s5": "Triglyceriden",23 "s6": "Bloedsuiker (glucose)",24 "target": "Doelscore (progressie)",25}26LABEL_TO_KEY = {v: k for k, v in FEATURE_LABELS.items()}27 28MEDICAL_MD = """29### Medisch nut 30 31**Wat zien we hier?** 32Ik heb een bestaande, anonieme gezondheidsdataset gebruikt die speciaal beschikbaar is gemaakt voor onderzoek en studie. In deze gegevens staan metingen van een grote groep patiënten, zoals **bloedwaarden, BMI, cholesterol en bloedsuiker**. 33 34Zo'n enorme berg cijfers is voor artsen en ziekenhuizen bijna niet in één keer te overzien. Het is gewoon te veel om met het blote oog patronen uit te halen. 35 36**Daar komt kunstmatige intelligentie om de hoek kijken.** 37Met deze techniek (PCA) kan de computer de data slim samenvatten en patronen zichtbaar maken. Dit programma dat ik heb ontworpen laat live zien hoe die samenvatting werkt. 38 39- Elke punt is één patiënt. 40- De kleur laat zien hoe hoog of laag een bepaalde meting is (standaard: BMI). 41- De pijlen (in de 2D-biplot) laten zien welke metingen het meeste invloed hebben. 42- Links bovenin kun je kiezen welke meting je als uitgangspunt wilt nemen. 43 44**En wat heb je hieraan?** 45In de praktijk gebruiken artsen en onderzoekers zo'n plot om patronen en verbanden te ontdekken. 👉 Het is dus niet alleen een mooi plaatje, maar echt een manier om grote hoeveelheden data sneller en slimmer te begrijpen. 46 47Met AI kunnen we patronen vinden die je met het blote oog nooit zou zien. Dat maakt dit niet alleen een mooie visualisatie, maar ook een knap stukje technologie met échte waarde voor onderzoek en zorg. 48 49**Speel zelf de onderzoeker!** 50Doe alsof je een arts bent en kies links bovenin een waarde, bijvoorbeeld **cholesterol**, **leeftijd** of **geslacht**. Klik daarna op **Update visualisaties** en ontdek je eigen patronen in de data. 51"""52 53# -------------------- Data helpers --------------------54def load_diabetes_df():55 d = datasets.load_diabetes()56 X = pd.DataFrame(d.data, columns=d.feature_names) # gestandaardiseerd57 y = pd.Series(d.target, name="target")58 df = X.copy(); df["target"] = y59 return df60 61def compute_overview_table(df: pd.DataFrame):62 keys = ["bmi","bp","s1","s2","s3","s4","s5","s6"]63 rows = []64 for k in keys:65 vals = df[k].dropna().values66 mean = float(vals.mean())67 pct_above = float((vals > 0).mean() * 100.0) # 0 ≈ globaal gemiddelde68 pct_below = float((vals < 0).mean() * 100.0)69 rows.append({70 "Meting": FEATURE_LABELS.get(k, k),71 "Gemiddelde (gestandaardiseerd)": round(mean, 3),72 "% boven gemiddelde": round(pct_above, 1),73 "% onder gemiddelde": round(pct_below, 1),74 })75 table = pd.DataFrame(rows)76 note = ("Let op: waarden in deze dataset zijn **gestandaardiseerd**. `0` ≈ algemeen gemiddelde. "77 "Positief = hoger dan gemiddeld, negatief = lager dan gemiddeld.")78 return table, note79 80# -------------------- PCA helpers --------------------81def compute_pca(df: pd.DataFrame, n_components: int, standardize: bool):82 feats = [c for c in df.columns if c != "target"]83 X = df[feats].values84 if standardize:85 scaler = StandardScaler(with_mean=True, with_std=True)86 Xs = scaler.fit_transform(X)87 else:88 Xs = X89 pca = PCA(n_components=min(int(n_components), Xs.shape[1]))90 Z = pca.fit_transform(Xs)91 loadings = pca.components_.T92 expl = pca.explained_variance_ratio_93 return feats, Xs, Z, loadings, expl94 95# -------------------- Plot builders --------------------96def build_biplot_plotly(df, Z, loadings, feats, color_key, arrow_scale=2.0):97 # Hover info98 fields = ["bmi","bp","s1","s2","s3","s4","s5","s6","age","sex","target"]99 hover_text = [100 "<br>".join(f"{FEATURE_LABELS.get(k,k)}: {df.iloc[i][k]:.3f}" for k in fields)101 for i in range(len(df))102 ]103 fig = go.Figure()104 fig.add_trace(go.Scatter(105 x=Z[:,0], y=Z[:,1], mode="markers",106 marker=dict(size=8, color=df[color_key].values),107 text=hover_text, hovertemplate="%{text}<extra></extra>"108 ))109 # loading pijlen110 for i, key in enumerate(feats):111 x = loadings[i,0]*arrow_scale; y = loadings[i,1]*arrow_scale112 fig.add_annotation(x=x, y=y, ax=0, ay=0, xref="x", yref="y", axref="x", ayref="y",113 showarrow=True, arrowhead=3)114 fig.add_annotation(x=x*1.05, y=y*1.05, text=FEATURE_LABELS.get(key,key),115 showarrow=False, font=dict(size=10))116 fig.update_layout(title="PCA-biplot (2D, hover)", xaxis_title="PC1", yaxis_title="PC2",117 margin=dict(l=10, r=10, t=40, b=10))118 return fig119 120def build_biplot_matplotlib(df, Z, loadings, feats, color_key, arrow_scale=2.0, point_size=32, alpha=0.85):121 fig = plt.figure()122 ax = fig.add_subplot(111)123 sc = ax.scatter(Z[:,0], Z[:,1], c=df[color_key].values, s=point_size, alpha=alpha)124 cbar = plt.colorbar(sc, ax=ax, pad=0.02); cbar.set_label(f"Kleur: {FEATURE_LABELS.get(color_key,color_key)}")125 ax.set_xlabel("PC1"); ax.set_ylabel("PC2"); ax.set_title("PCA-biplot — PNG-export")126 for i,key in enumerate(feats):127 x=loadings[i,0]*arrow_scale; y=loadings[i,1]*arrow_scale128 ax.arrow(0,0,x,y, head_width=0.05, head_length=0.08, fc="k", ec="k", length_includes_head=True)129 ax.text(x*1.08, y*1.08, FEATURE_LABELS.get(key,key), fontsize=9, ha="center", va="center")130 ax.axhline(0,color="grey",linewidth=0.6,linestyle=":"); ax.axvline(0,color="grey",linewidth=0.6,linestyle=":")131 ax.grid(True,linestyle=":",linewidth=0.6); fig.tight_layout()132 return fig133 134def build_pca3d(Z3, color_vals):135 fig = go.Figure(data=[go.Scatter3d(x=Z3[:,0], y=Z3[:,1], z=Z3[:,2], mode="markers",136 marker=dict(size=4, color=color_vals, opacity=0.85))])137 fig.update_layout(title="PCA 3D — PC1·PC2·PC3 (sleep om te draaien)",138 scene=dict(xaxis_title="PC1", yaxis_title="PC2", zaxis_title="PC3"),139 margin=dict(l=10, r=10, t=40, b=10))140 return fig141 142def build_variance_plot(expl):143 fig = plt.figure()144 ax = fig.add_subplot(111)145 xs = np.arange(1, len(expl)+1)146 ax.bar(xs, expl, width=0.8, align="center")147 ax.plot(xs, np.cumsum(expl), marker="o")148 ax.set_xticks(xs); ax.set_xlabel("Principal Component"); ax.set_ylabel("Explained variance ratio")149 ax.set_title("Uitlegvariantie per component (balken) + cumulatief (lijn)")150 ax.grid(True, linestyle=":", linewidth=0.6); fig.tight_layout()151 return fig152 153def build_hist_box(df: pd.DataFrame, color_key: str):154 series = df[color_key].dropna()155 label = FEATURE_LABELS.get(color_key, color_key)156 fig_hist = px.histogram(x=series, nbins=30, title=f"Histogram — {label}", labels={"x": label})157 fig_hist.update_layout(xaxis_title=label, yaxis_title="Aantal", margin=dict(l=10, r=10, t=40, b=10))158 fig_box = px.box(y=series, points="outliers", title=f"Boxplot — {label}", labels={"y": label})159 fig_box.update_layout(yaxis_title=label, margin=dict(l=10, r=10, t=40, b=10))160 return fig_hist, fig_box161 162# -------------------- Controllers --------------------163def controller(color_label="BMI (Body Mass Index)", n_components=10, standardize=True, arrow_scale=2.0):164 df = load_diabetes_df()165 feats, Xs, Z, loadings, expl = compute_pca(df, n_components, standardize)166 color_key = LABEL_TO_KEY.get(color_label, "bmi")167 color_vals = df[color_key].values168 169 fig_biplot = build_biplot_plotly(df, Z, loadings, feats, color_key, arrow_scale=arrow_scale)170 if Z.shape[1] < 3:171 pca3 = PCA(n_components=3); Z3 = pca3.fit_transform(Xs)172 else:173 Z3 = Z[:, :3]174 fig3d = build_pca3d(Z3, color_vals)175 fig_variance = build_variance_plot(expl)176 fig_hist, fig_box = build_hist_box(df, color_key)177 178 load_df = pd.DataFrame({179 "feature_key": feats,180 "PC1_loading": loadings[:, 0],181 "PC2_loading": loadings[:, 1],182 "PC1_abs": np.abs(loadings[:, 0]),183 "PC2_abs": np.abs(loadings[:, 1]),184 })185 load_df["Feature (PC1)"] = load_df["feature_key"].map(lambda k: FEATURE_LABELS.get(k, k))186 load_df["Feature (PC2)"] = load_df["feature_key"].map(lambda k: FEATURE_LABELS.get(k, k))187 top_pc1 = load_df.sort_values("PC1_abs", ascending=False)[["Feature (PC1)", "PC1_loading"]].head(6).reset_index(drop=True)188 top_pc2 = load_df.sort_values("PC2_abs", ascending=False)[["Feature (PC2)", "PC2_loading"]].head(6).reset_index(drop=True)189 max_len = max(len(top_pc1), len(top_pc2))190 top_pc1 = top_pc1.reindex(range(max_len)); top_pc2 = top_pc2.reindex(range(max_len))191 table = pd.concat([top_pc1, top_pc2], axis=1)192 193 overview_df, overview_note = compute_overview_table(df)194 195 summary_md = f"""196### Wat zie je hier?197- **Klik op _Update visualisaties_** om alles te verversen met jouw keuze.198- **Hover** over punten voor exacte waarden (BMI, bloeddruk, cholesterol, glucose, leeftijd, geslacht, etc.).199- **2D-biplot** met pijlen (belangrijkste metingen) en **3D-view** voor extra diepte.200- **Uitlegvariantieplot**: laat zien hoeveel variatie elke component uitlegt.201- **Histogram + boxplot**: verdeling en spreiding van de gekozen meting ({FEATURE_LABELS.get(color_key,color_key)}).202"""203 return fig_biplot, fig3d, fig_variance, table, overview_df, overview_note, summary_md204 205def animate_pca(color_label="BMI (Body Mass Index)", point_size=32, alpha=0.85, n_components=10, standardize=True, frames=40, pause=0.0):206 df = load_diabetes_df()207 feats, Xs, Z, loadings, expl = compute_pca(df, n_components, standardize)208 color_key = LABEL_TO_KEY.get(color_label, "bmi")209 color_vals = df[color_key].values210 for i in range(frames):211 t = i / max(1, frames-1)212 w1 = min(1.0, t * 2.0); w2 = max(0.0, (t - 0.5) * 2.0)213 coords = np.column_stack([Z[:, 0] * w1, Z[:, 1] * w2])214 fig = plt.figure()215 ax = fig.add_subplot(111)216 ax.scatter(coords[:, 0], coords[:, 1], c=color_vals, s=point_size, alpha=alpha)217 ax.set_xlabel("PC1 (opbouw)"); ax.set_ylabel("PC2 (opbouw)")218 title = "PCA-projectie (animatie) — " + ("PC1 →" if w2 == 0 else "PC1 + PC2")219 ax.set_title(f"{title} — frame {i+1}/{frames}")220 ax.axhline(0, color="grey", linewidth=0.6, linestyle=":"); ax.axvline(0, color="grey", linewidth=0.6, linestyle=":")221 ax.grid(True, linestyle=":", linewidth=0.6); fig.tight_layout()222 yield fig223 if pause > 0:224 time.sleep(pause)225 226def export_biplot_png(color_label="BMI (Body Mass Index)", arrow_scale=2.0, point_size=32, alpha=0.85, n_components=10, standardize=True):227 df = load_diabetes_df()228 feats, Xs, Z, loadings, expl = compute_pca(df, n_components, standardize)229 color_key = LABEL_TO_KEY.get(color_label, "bmi")230 fig = build_biplot_matplotlib(df, Z, loadings, feats, color_key, arrow_scale=arrow_scale, point_size=point_size, alpha=alpha)231 path = f"/mnt/data/biplot_{int(time.time())}.png"232 fig.savefig(path, dpi=150, bbox_inches="tight"); plt.close(fig)233 return path234 235def export_variance_png(n_components=10, standardize=True):236 df = load_diabetes_df()237 feats, Xs, Z, loadings, expl = compute_pca(df, n_components, standardize)238 fig = build_variance_plot(expl)239 path = f"/mnt/data/variance_{int(time.time())}.png"240 fig.savefig(path, dpi=150, bbox_inches="tight"); plt.close(fig)241 return path242 243# -------------------- UI --------------------244with gr.Blocks(title="PCA Dashboard — Diabetes (netjes & compleet)") as demo:245 gr.HTML("""246 <style>247 .callout {padding:12px 14px; border-left:4px solid #2563eb; background:#f1f5f9; border-radius:8px; margin: 8px 0 18px;}248 .cta {padding:10px 12px; border:1px dashed #2563eb; background:#eff6ff; border-radius:8px; margin-top:6px;}249 </style>250 """)251 252 gr.Markdown("# PCA Dashboard — Diabetes (netjes & compleet)")253 gr.Markdown(MEDICAL_MD)254 gr.HTML('<div class="callout"><b>Belangrijk:</b> kies links je instellingen en klik daarna op <b>Update visualisaties</b>. Wil je de stap-voor-stap projectie zien? Klik op <b>▶ Animate PCA</b>.</div>')255 256 with gr.Row():257 with gr.Column(scale=1):258 with gr.Group():259 gr.Markdown("### Instellingen")260 color_choices = [FEATURE_LABELS[k] for k in ["bmi","bp","s1","s2","s3","s4","s5","s6","age","sex","target"]]261 color_feat = gr.Dropdown(choices=color_choices, value=FEATURE_LABELS["bmi"], label="Kleur op meting")262 n_components = gr.Slider(3, 10, value=10, step=1, label="Aantal PCA-componenten")263 standardize = gr.Checkbox(value=True, label="Standaardiseer metingen (aanbevolen)")264 arrow_scale = gr.Slider(0.5, 5.0, value=2.0, step=0.1, label="Pijl-schaal (2D-biplot)")265 run_btn = gr.Button("🔄 Update visualisaties")266 gr.HTML('<div class="cta"><b>Klik hierna op: "🔄 Update visualisaties"</b> om alle grafieken te verversen.</div>')267 with gr.Group():268 gr.Markdown("### Animatie")269 animate_btn = gr.Button("▶ Animate PCA (PC1 → PC2)")270 gr.HTML('<div class="cta"><b>Klik op: "▶ Animate PCA"</b> om de projectie stap-voor-stap te zien.</div>')271 anim_plot = gr.Plot(label="Animatie van projectie")272 with gr.Group():273 gr.Markdown("### Downloads")274 dl_biplot = gr.DownloadButton("Download biplot (PNG)")275 dl_var = gr.DownloadButton("Download variatieplot (PNG)")276 277 with gr.Column(scale=2):278 with gr.Row():279 with gr.Column():280 gr.Markdown("### Biplot (2D, hover)")281 plot_biplot = gr.Plotly()282 with gr.Column():283 gr.Markdown("### 3D PCA (PC1–PC3)")284 plot3d = gr.Plotly()285 with gr.Row():286 with gr.Column():287 gr.Markdown("### Uitlegvariantie")288 plot_expl = gr.Plot()289 with gr.Column():290 gr.Markdown("### Top-features (PC1 / PC2)")291 table = gr.Dataframe(headers=["Feature (PC1)", "Loading PC1", "Feature (PC2)", "Loading PC2"], row_count=6)292 with gr.Row():293 with gr.Column():294 gr.Markdown("### Histogram")295 plot_hist = gr.Plotly()296 with gr.Column():297 gr.Markdown("### Boxplot")298 plot_box = gr.Plotly()299 with gr.Row():300 with gr.Column():301 gr.Markdown("### Overzicht (gemiddelden & verdeling)")302 overview_tbl = gr.Dataframe(interactive=False)303 with gr.Column():304 gr.Markdown("### Samenvatting")305 summary = gr.Markdown()306 overview_note_md = gr.Markdown()307 308 inputs = [color_feat, n_components, standardize, arrow_scale]309 run_btn.click(fn=controller, inputs=inputs,310 outputs=[plot_biplot, plot3d, plot_expl, table, overview_tbl, overview_note_md, summary])311 demo.load(fn=controller, inputs=inputs,312 outputs=[plot_biplot, plot3d, plot_expl, table, overview_tbl, overview_note_md, summary])313 314 animate_btn.click(fn=animate_pca, inputs=[color_feat], outputs=anim_plot)315 316 dl_biplot.click(fn=export_biplot_png, inputs=[color_feat, arrow_scale], outputs=[dl_biplot])317 dl_var.click(fn=export_variance_png, inputs=[], outputs=[dl_var])318 319if __name__ == "__main__":320 demo.queue().launch(server_name="0.0.0.0", server_port=7860, ssr_mode=False, show_api=False)321