Upload 2 files

app.py

@@ -4,7 +4,6 @@ import numpy as np
 import matplotlib.pyplot as plt
 from sklearn import datasets
 from sklearn.utils import shuffle
-import csv, os, tempfile
 
 EXPLAIN_MD = """
 ### Wat testen we hier?
@@ -79,8 +78,6 @@ def sgd_train_generator(lr, epochs, batch_size, seed, split_seed):
 
     rng = np.random.RandomState(int(seed))
 
-    state_for_download = None
-
     for epoch in range(1, int(epochs) + 1):
         # shuffle train set
         x_tr, y_tr = shuffle(x_tr, y_tr, random_state=rng)
@@ -147,51 +144,9 @@ def sgd_train_generator(lr, epochs, batch_size, seed, split_seed):
             f"{CONCLUSION_MD}"
         )
 
-        # Bewaar state voor download (laatste epoch)
-        state_for_download = {
-            "x_test": x_te,
-            "y_test": y_te,
-            "y_pred": y_te_pred,
-            "w": w,
-            "b": b,
-            "mse_train": mse_tr,
-            "mse_test": mse_te,
-            "r2_test": r2_te,
-        }
-
-        yield fig_main, fig_loss, summary, state_for_download
-
-def prepare_download(state):
-    if not state:
-        return None
-    # Schrijf CSV met test set en voorspellingen
-    fd, path = tempfile.mkstemp(suffix="_diabetes_bmi_results.csv")
-    os.close(fd)
-    with open(path, "w", newline="", encoding="utf-8") as f:
-        writer = csv.writer(f)
-        writer.writerow(["bmi_normalized", "y_true", "y_pred", "residual"]) 
-        for x, yt, yp in zip(state["x_test"], state["y_test"], state["y_pred"]):
-            writer.writerow([float(x), float(yt), float(yp), float(yt-yp)])
-        # Voeg onderaan een lege regel + metrics toe
-        writer.writerow([])
-        writer.writerow(["w", state["w"]])
-        writer.writerow(["b", state["b"]])
-        writer.writerow(["mse_train", state["mse_train"]])
-        writer.writerow(["mse_test", state["mse_test"]])
-        writer.writerow(["r2_test", state["r2_test"]])
-    return path
+        yield fig_main, fig_loss, summary
 
 with gr.Blocks(title="Diabetes: BMI → Progressiescore (Live Regressie)") as demo:
-    # Custom CSS to color the buttons (using elem_id selectors)
-    gr.HTML("""
-    <style>
-      #train-btn button { background:#2563eb; color:white; border:none; }
-      #train-btn button:hover { filter: brightness(0.95); }
-      #download-btn button { background:#059669; color:white; border:none; }
-      #download-btn button:hover { filter: brightness(0.95); }
-    </style>
-    """)
-
     gr.Markdown("# Diabetes: BMI → Progressiescore (Live Lineaire Regressie)")
     gr.Markdown(EXPLAIN_MD)
 
@@ -202,35 +157,27 @@ with gr.Blocks(title="Diabetes: BMI → Progressiescore (Live Regressie)") as de
             batch = gr.Slider(8, 256, value=64, step=1, label="Batchgrootte")
             seed = gr.Slider(0, 9999, value=42, step=1, label="Training seed")
             split_seed = gr.Slider(0, 9999, value=7, step=1, label="Train/test split seed")
-            train_btn = gr.Button("Train live", elem_id="train-btn", variant="primary")
-            download_btn = gr.DownloadButton(
-                label="Download resultaten (CSV)", elem_id="download-btn", file_name="diabetes_bmi_results.csv"
-            )
-            # Story direct onder de knoppen
+            train_btn = gr.Button("Train live")
+            # Story direct onder de knop
             gr.Markdown(STORY_MD)
         with gr.Column(scale=2):
             plot_main = gr.Plot(label="Data (train/test) & regressielijn (live)")
             plot_loss = gr.Plot(label="Loss-curve (MSE per epoch) — train vs test")
             results = gr.Markdown()
 
-    results_state = gr.State()
-
-    # Training starten via knop (streaming)
+    # Training starten via knop
     train_btn.click(
         fn=sgd_train_generator,
         inputs=[lr, epochs, batch, seed, split_seed],
-        outputs=[plot_main, plot_loss, results, results_state]
+        outputs=[plot_main, plot_loss, results]
     )
 
     # Auto-train bij laden met default-waarden
     demo.load(
         fn=sgd_train_generator,
         inputs=[lr, epochs, batch, seed, split_seed],
-        outputs=[plot_main, plot_loss, results, results_state]
+        outputs=[plot_main, plot_loss, results]
     )
 
-    # Download-button: maak CSV vanuit state
-    download_btn.click(fn=prepare_download, inputs=[results_state], outputs=download_btn)
-
 if __name__ == "__main__":
     demo.launch()