Update app.py

app.py

@@ -1,14 +1,12 @@
 import gradio as gr
 import pandas as pd
 import numpy as np
-from sklearn.model_selection import train_test_split, GridSearchCV
-from sklearn.svm import SVC
 from sklearn.preprocessing import StandardScaler
 import scipy
 from scipy import signal
 import pickle
-import asyncio
 
+# Global variable to store the uploaded data
 global_data = None
 
 def get_data_preview(file):
@@ -18,37 +16,29 @@ def get_data_preview(file):
 
 def label_data(ranges):
     global global_data
-    for start, end, label in ranges.values:
+    for start, end, label in ranges:
         global_data.loc[start:end, 'label'] = label
-    labeled_data = pd.concat([global_data.head(), global_data.tail()])
-    return labeled_data
-
-
-#def label_data(ranges):
-    #global global_data
-    #for start, end, label in ranges.values:
-    #    global_data.loc[start:end, 'label'] = label
-    #return global_data
-
-
-def preprocess_data(data):
-    data.drop(columns=data.columns[0], axis=1, inplace=True)
-    data.columns = ['raw_eeg', 'label']
-    raw_data = data['raw_eeg']
-    labels_old = data['label']
+    return global_data.head()
 
+def preprocess_data():
+    global global_data
+    global_data.drop(columns=global_data.columns[0], axis=1, inplace=True)
+    global_data.columns = ['raw_eeg', 'label']
+    raw_data = global_data['raw_eeg']
+    labels_old = global_data['label']
+    
     sampling_rate = 512
     notch_freq = 50.0
     lowcut, highcut = 0.5, 30.0
-
+    
     nyquist = (0.5 * sampling_rate)
     notch_freq_normalized = notch_freq / nyquist
     b_notch, a_notch = signal.iirnotch(notch_freq_normalized, Q=0.05, fs=sampling_rate)
-
+    
     lowcut_normalized = lowcut / nyquist
     highcut_normalized = highcut / nyquist
     b_bandpass, a_bandpass = signal.butter(4, [lowcut_normalized, highcut_normalized], btype='band')
-
+    
     features = []
     labels = []
 
@@ -94,56 +84,38 @@ def preprocess_data(data):
         segment_features = {**segment_features, **additional_features}
         features.append(segment_features)
         labels.append(labels_old[i])
-
+    
     columns = ['E_alpha', 'E_beta', 'E_theta', 'E_delta', 'alpha_beta_ratio', 'peak_frequency', 'spectral_centroid', 'spectral_slope']
     df_features = pd.DataFrame(features, columns=columns)
     df_features['label'] = labels
-    return df_features
-
-def train_model():
-    global global_data
-    data = preprocess_data(global_data)
+    
     scaler = StandardScaler()
-    X = data.drop('label', axis=1)
-    y = data['label']
-    X_scaled = scaler.fit_transform(X)
-    X_train, X_test, y_train, y_test = train_test_split(X_scaled, y, test_size=0.2, random_state=42)
+    X_scaled = scaler.fit_transform(df_features.drop('label', axis=1))
+    df_scaled = pd.DataFrame(X_scaled, columns=columns)
+    df_scaled['label'] = df_features['label']
     
-    param_grid = {'C': [0.1, 1, 10, 100], 'gamma': ['scale', 'auto', 0.1, 0.01, 0.001, 0.0001], 'kernel': ['rbf']}
-    svc = SVC(probability=True)
-    grid_search = GridSearchCV(estimator=svc, param_grid=param_grid, cv=5, verbose=2, n_jobs=-1)
-    grid_search.fit(X_train, y_train)
+    processed_data_filename = 'processed_data.csv'
+    df_scaled.to_csv(processed_data_filename, index=False)
     
-    model = grid_search.best_estimator_
-    model_filename = 'model.pkl'
     scaler_filename = 'scaler.pkl'
-    
-    with open(model_filename, 'wb') as file:
-        pickle.dump(model, file)
-    
     with open(scaler_filename, 'wb') as file:
         pickle.dump(scaler, file)
     
-    return "Training complete! Model and scaler saved.", model_filename, scaler_filename
-
+    return "Data preprocessing complete! Download the processed data and scaler below.", processed_data_filename, scaler_filename
 
 with gr.Blocks() as demo:
     file_input = gr.File(label="Upload CSV File")
     data_preview = gr.Dataframe(label="Data Preview", interactive=False)
     ranges_input = gr.Dataframe(headers=["Start Index", "End Index", "Label"], label="Ranges for Labeling")
-    
-    #start_input = gr.Number(label="Start Index", value=0)
-    #end_input = gr.Number(label="End Index", value=100)
-    #label_input = gr.Number(label="Label Value", value=1)
     labeled_data_preview = gr.Dataframe(label="Labeled Data Preview", interactive=False)
-    training_status = gr.Textbox(label="Training Status")
-    model_file = gr.File(label="Download Trained Model")
+    preprocessing_status = gr.Textbox(label="Preprocessing Status")
+    processed_data_file = gr.File(label="Download Processed Data")
     scaler_file = gr.File(label="Download Scaler")
-
+    
     file_input.upload(get_data_preview, inputs=file_input, outputs=data_preview)
     label_button = gr.Button("Label Data")
-    label_button.click(label_data, inputs=[ranges_input], outputs=labeled_data_preview, queue=True)
-    train_button = gr.Button("Train Model")
-    train_button.click(train_model, outputs=[training_status, model_file, scaler_file])
+    label_button.click(label_data, inputs=[ranges_input], outputs=labeled_data_preview)
+    preprocess_button = gr.Button("Preprocess Data")
+    preprocess_button.click(preprocess_data, outputs=[preprocessing_status, processed_data_file, scaler_file])
 
-demo.launch()
+demo.launch()