Finalize up the AI Trainer

WebServer/AIPython/ai-trainer.py

@@ -7,16 +7,23 @@ os.environ["CUDA_VISIBLE_DEVICES"] = "-1"
 from sklearn.model_selection import train_test_split
 from keras import Sequential, layers, optimizers, losses
 
-def TrainAI():
-    # Pull New Data
-    datapuller.pull()
+def TrainAI(include_pull):
 
-    # Get the CWD for pathing due to being called from C# now
+    if (include_pull):
+        # Pull New Data
+        datapuller.pull()
+
+    # Get the CWD for pathing due to being called from C#
     SCRIPT_DIR = os.path.dirname(os.path.abspath(__file__))
     DATA_DIR = os.path.join(SCRIPT_DIR, "data")
 
     # Load the dataset
     dataset = pd.read_parquet(os.path.join(DATA_DIR, "stocks.parquet"))
+
+    # Replace Infinity with 0 -> This fixes the AI mental breakdown
+    dataset['Volume_Chg'] = dataset['Volume_Chg'].replace([np.inf, -np.inf], 0)
+
+    # Remove indicators and set the target
     X = dataset.drop('Target_Close', axis=1)
     X = dataset.drop('Target_Direction', axis=1)
     Y = dataset['Target_Close']
@@ -31,8 +38,8 @@ def TrainAI():
     normalizer = layers.Normalization(axis=-1)
     normalizer.adapt(np.array(train_features))
 
-    # Start with a linear model
-    dnn_linear_model = Sequential([
+    # Create the DNN
+    dnn_model = Sequential([
         layers.Input(shape=(train_features.shape[1],)), # Load the feature count dynamically
         normalizer,
         layers.Dense(64, activation='elu'),
@@ -41,28 +48,26 @@ def TrainAI():
     ])
 
     # Configure the model
-    dnn_linear_model.compile(
-        optimizer=optimizers.Adam(learning_rate=0.0001, clipvalue=1.0),
+    dnn_model.compile(
+        optimizer=optimizers.Adam(learning_rate=0.00001, clipvalue=1.0),
         loss=losses.Huber()
     )
 
     # Show the summary before training the model
-    dnn_linear_model.summary()
+    dnn_model.summary()
 
     # Train the model
-    Training_Data = dnn_linear_model.fit( 
+    Training_Data = dnn_model.fit( 
         train_features,
         train_labels,
         batch_size=64,
-        epochs=100,
-        # Show progress
-        verbose=1,
-        # Calculate validation results on 20% of the training data.
-        validation_split = 0.2
+        epochs=39,                  # Tuned to the point before overfitting
+        verbose=1,                  # Show progress
+        validation_split = 0.2      # Calculate validation results on 20% of the training data.
     )
 
     # Predict
-    test_predictions = dnn_linear_model.predict(test_features).flatten()
+    test_predictions = dnn_model.predict(test_features).flatten()
     a = plt.axes(aspect='equal')
     plt.scatter(test_labels, test_predictions)
     plt.xlabel('True Values')
@@ -73,11 +78,14 @@ def TrainAI():
     _ = plt.plot(lims, lims)
 
 
-    test_results = dnn_linear_model.evaluate(
+    # Current Test Results: 1.221876300405711e-05
+    test_results = dnn_model.evaluate(
         test_features, test_labels, verbose=0
     )
-
     print(f"Test Results: {test_results}")
 
     # Save the model
-    dnn_linear_model.save(os.path.join(DATA_DIR, "model.keras"))
+    dnn_model.save(os.path.join(DATA_DIR, "model.keras"))
+
+if __name__ == "__main__":
+    TrainAI(False)