init

old/datasets.py

@@ -0,0 +1,15 @@
+import os
+import subprocess
+
+class Datasets:
+    @staticmethod
+    def download_kaggle_dataset(dataset_id, download_path):
+        os.environ["KAGGLE_CONFIG_DIR"] = os.path.expanduser("~/.kaggle")
+        command = ["kaggle", "datasets", "download", "-d", dataset_id, "-p", download_path]
+
+        try:
+            subprocess.run(command, check=True)
+            print(f"Dataset downloaded successfully to {download_path}")
+        except subprocess.CalledProcessError as e:
+            print(f"Error downloading dataset: {e}")
+

old/download_dataset.py

@@ -0,0 +1,5 @@
+from datasets import Datasets
+
+dataset_id = "mczielinski/bitcoin-historical-data"
+download_path = "./data"
+Datasets.download_kaggle_dataset(dataset_id, download_path)

old/drop_data.py

@@ -0,0 +1,36 @@
+import sqlite3
+from datetime import datetime
+
+# Specify the database file path
+db_path = 'bitcoin_historical_data.db'
+
+# Create a connection to the database
+connection = sqlite3.connect(db_path)
+
+# Create a cursor object
+cursor = connection.cursor()
+
+# Define the date threshold
+date_threshold = datetime(2025, 1, 15)
+
+# Convert the date threshold to the format used in SQLite (YYYY-MM-DD HH:MM:SS.SSS)
+date_threshold_str = date_threshold.strftime('%Y-%m-%d 00:00:00.000')
+
+# SQL query to delete rows with Timestamp greater than the date threshold
+query = """
+DELETE FROM bitcoin_data
+WHERE Timestamp > ?
+"""
+
+# Execute the query with the date threshold as a parameter
+cursor.execute(query, (date_threshold_str,))
+
+# Commit the changes
+connection.commit()
+
+# Get the number of deleted rows
+deleted_rows = cursor.rowcount
+print(f"Deleted {deleted_rows} rows with Timestamp greater than January 15th, 2025")
+
+# Close the connection
+connection.close()

old/new_data_prediction.py

@@ -0,0 +1,21 @@
+from BitcoinPricePredictor import BitcoinPricePredictor
+from tensorflow.keras.models import load_model
+from sklearn.metrics import confusion_matrix
+
+if __name__ == "__main__":
+    model = load_model('models/model_2025-01-21_04-49-43.h5')
+    predictor = BitcoinPricePredictor(model=model, db_path='bitcoin_historical_data.db')
+
+    missing_data = predictor.load_new_data_from_model()
+
+    print(f"missing data {len(missing_data)}")
+    if not missing_data.empty:
+        predictions, reality = predictor.make_predictions_w_reality(missing_data)
+        print(f"predictions {len(predictions)}")
+
+        cm = confusion_matrix(reality, predictions[1:])
+        print("Confusion Matrix:")
+        print(cm)
+    else:
+        print("No new data found.")
+

old/parse_btc_csv_to_db.py

@@ -0,0 +1,32 @@
+import pandas as pd
+from sqlalchemy import create_engine, text
+
+# Load the dataset
+df = pd.read_csv('./data/btcusd_1-min_data.csv')
+
+# Preprocess the data
+df['Timestamp'] = pd.to_datetime(df['Timestamp'], unit='s')
+df.set_index('Timestamp', inplace=True)
+
+# Remove rows with invalid Timestamps
+df = df[~df.index.isna()]
+
+# Create a connection to the SQLite database
+engine = create_engine('sqlite:///bitcoin_historical_data.db')
+
+# Check if the table already exists and get the last timestamp from the database
+with engine.connect() as connection:
+    query = text("SELECT MAX(Timestamp) FROM bitcoin_data")
+    last_timestamp = connection.execute(query).fetchone()[0]
+
+# If there is no data in the table, last_timestamp will be None
+if last_timestamp is not None:
+    # Filter the new data to include only rows with a timestamp later than the last timestamp in the database
+    df = df[df.index > last_timestamp]
+
+# If there are new rows, append them to the database
+if not df.empty:
+    df.to_sql('bitcoin_data', engine, if_exists='append', index=True)
+    print(f"Added {len(df)} new rows to the database.")
+else:
+    print("No new data to add.")