upload the ai trainer

WebServer/AIPython/ai-trainer.py

@@ -1,49 +1,54 @@
 import pandas as pd
 import numpy as np
 import matplotlib.pyplot as plt
+import datapuller
 import os
 from sklearn.model_selection import train_test_split
 from keras import Sequential, layers, optimizers
 
-# Get the CWD for pathing due to being called from C# now
+def TrainAI():
-SCRIPT_DIR = os.path.dirname(os.path.abspath(__file__))
+    # Pull New Data
-DATA_DIR = os.path.join(SCRIPT_DIR, "data")
+    datapuller.pull()
 
-# Load the dataset
+    # Get the CWD for pathing due to being called from C# now
-dataset = pd.read_parquet(os.path.join(DATA_DIR, "stocks.parquet"))
+    SCRIPT_DIR = os.path.dirname(os.path.abspath(__file__))
-X = dataset.drop('Volatility_5', axis=1)
+    DATA_DIR = os.path.join(SCRIPT_DIR, "data")
-Y = dataset['Volatility_5']
 
-# Show the datatypes
+    # Load the dataset
-print(dataset.dtypes)
+    dataset = pd.read_parquet(os.path.join(DATA_DIR, "stocks.parquet"))
+    X = dataset.drop('Volatility_5', axis=1)
+    Y = dataset['Volatility_5']
 
-# Split out the test and train
+    # Show the datatypes
-train_features, test_features, train_labels, test_labels = train_test_split(X, Y, test_size=0.2)
+    print(dataset.dtypes)
 
-# Create a normalizer to nomralize the data
+    # Split out the test and train
-normalizer = layers.Normalization(axis=-1)
+    train_features, test_features, train_labels, test_labels = train_test_split(X, Y, test_size=0.2)
-normalizer.adapt(np.array(train_features))
 
-# Start with a linear model
+    # Create a normalizer to nomralize the data
-dnn_linear_model = Sequential([
+    normalizer = layers.Normalization(axis=-1)
+    normalizer.adapt(np.array(train_features))
+
+    # Start with a linear model
+    dnn_linear_model = Sequential([
         layers.Input(shape=(9,)), # Explicitly tell Keras there are 9 features
         normalizer,
         layers.Dense(64, activation='relu'),
         layers.Dense(64, activation='relu'),
         layers.Dense(1)
-])
+    ])
 
-# Configure the model
+    # Configure the model
-dnn_linear_model.compile(
+    dnn_linear_model.compile(
         optimizer=optimizers.Adam(learning_rate=0.001),
         loss='mean_absolute_error'
-)
+    )
 
-# Show the summary before training the model
+    # Show the summary before training the model
-dnn_linear_model.summary()
+    dnn_linear_model.summary()
 
-# Train the model
+    # Train the model
-Training_Data = dnn_linear_model.fit( 
+    Training_Data = dnn_linear_model.fit( 
         train_features,
         train_labels,
         epochs=100,
@@ -51,23 +56,23 @@ Training_Data = dnn_linear_model.fit(
         verbose=1,
         # Calculate validation results on 20% of the training data.
         validation_split = 0.2
-)
+    )
 
-# Predict
+    # Predict
-test_predictions = dnn_linear_model.predict(test_features).flatten()
+    test_predictions = dnn_linear_model.predict(test_features).flatten()
-a = plt.axes(aspect='equal')
+    a = plt.axes(aspect='equal')
-plt.scatter(test_labels, test_predictions)
+    plt.scatter(test_labels, test_predictions)
-plt.xlabel('True Values')
+    plt.xlabel('True Values')
-plt.ylabel('Predictions')
+    plt.ylabel('Predictions')
-lims = [0, 50]
+    lims = [0, 50]
-plt.xlim(lims)
+    plt.xlim(lims)
-plt.ylim(lims)
+    plt.ylim(lims)
-_ = plt.plot(lims, lims)
+    _ = plt.plot(lims, lims)
 
 
-test_results = dnn_linear_model.evaluate(
+    test_results = dnn_linear_model.evaluate(
         test_features, test_labels, verbose=0
-)
+    )
 
-# Save the model
+    # Save the model
-dnn_linear_model.save(os.path.join(DATA_DIR, "model.keras"))
+    dnn_linear_model.save(os.path.join(DATA_DIR, "model.keras"))