Neural_Network_for_Finance.ipynb

{"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"provenance":[],"authorship_tag":"ABX9TyOCi+wggSGgALqnIfLe9Ump"},"kernelspec":{"name":"python3","display_name":"Python 3"},"language_info":{"name":"python"}},"cells":[{"cell_type":"markdown","source":["# Neural Network using YFinance data"],"metadata":{"id":"3TMF9z7HdQU9"}},{"cell_type":"code","source":["import yfinance as yf\n","\n","# Download stock data from Yahoo Finance\n","stock_symbol = 'INFY.NS'  # Replace with the stock symbol of your choice\n","stock_data = yf.download(stock_symbol, start='2010-01-01', end='2024-02-26')\n","\n","# Create a DataFrame from the downloaded data\n","df = pd.DataFrame(stock_data)\n","print(df)"],"metadata":{"id":"F3wpouYt7OqZ"},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":["## Using Ichimoku Cloud indicator"],"metadata":{"id":"-iC9Ly76dl3_"}},{"cell_type":"code","source":["import pandas as pd\n","import numpy as np\n","import matplotlib.pyplot as plt\n","from sklearn.preprocessing import MinMaxScaler\n","from sklearn.model_selection import train_test_split\n","from keras.models import Sequential\n","from keras.layers import Dense, LSTM, Dropout\n","from keras.optimizers import Adagrad\n","from keras.callbacks import EarlyStopping\n","\n","# Calculate Ichimoku Cloud indicator\n","df['Tenkan-sen'] = df['Close'].rolling(window=9).mean()\n","df['Kijun-sen'] = df['Close'].rolling(window=26).mean()\n","df['Senkou Span A'] = (df['Tenkan-sen'] + df['Kijun-sen']) / 2\n","df['Senkou Span B'] = (df['High'].rolling(window=52).max() + df['Low'].rolling(window=52).min()) / 2\n","df['Chikou Span'] = df['Close'].shift(-26)\n","df = df[:-26]  # drop the last 26 rows\n","df.fillna(0, inplace=True)\n","\n","# Create features for the neural network\n","features = df[['Tenkan-sen', 'Kijun-sen', 'Senkou Span A', 'Senkou Span B', 'Chikou Span']]\n","\n","print(df)"],"metadata":{"id":"nRbrUUWd01Va"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["# Scale the features\n","scaler = MinMaxScaler()\n","features_scaled = scaler.fit_transform(features)\n","\n","# Split the df into training and testing sets\n","X_train, X_test, y_train, y_test = train_test_split(features_scaled, df['Close'], test_size=0.2, random_state=108)\n","\n","# Create the neural network model\n","model = Sequential()\n","model.add(LSTM(100, input_shape=(X_train.shape[1], 1)))\n","model.add(Dropout(0.4)) # Dropout\n","model.add(Dense(1000)) # Dense\n","model.compile(loss='mean_squared_error', optimizer=Adagrad(learning_rate=0.0001))\n","\n","# Train the model\n","model.fit(X_train, y_train, epochs=100, batch_size=32, validation_data=(X_test, y_test), callbacks=[EarlyStopping(patience=10)])\n","\n","# Make predictions\n","predictions = model.predict(X_test)\n","\n","# Plot the predictions\n","plt.plot(y_test)\n","plt.plot(predictions)\n","plt.show()"],"metadata":{"id":"cpPeGTt87pki"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["y_test_values = y_test.values\n","print(y_test_values)"],"metadata":{"id":"BfG5D3IYC18i"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["y_pred = model.predict(X_test)\n","y_pred = y_pred.flatten()\n","y_pred = y_pred[:657]\n","\n","# Evaluate the model\n","mse = model.evaluate(X_test, y_test)\n","print(\"MSE: \", mse)\n","\n","# Calculate mean absolute error\n","mae = mean_absolute_error(y_test_values, y_pred)\n","print(\"MAE:\", mae)\n","\n","# Calculate mean squared logarithmic error\n","from sklearn.metrics import mean_squared_log_error\n","msle = mean_squared_log_error(y_test_values, y_pred)\n","print(\"MSLE:\", msle)\n","\n","# Calculate coefficient of determination (R-squared)\n","from sklearn.metrics import r2_score\n","r2 = r2_score(y_test_values, y_pred)\n","print(\"R-squared:\", r2)\n","\n","# Plot actual vs predicted values\n","import matplotlib.pyplot as plt\n","plt.plot(y_test_values)\n","plt.plot(y_pred)\n","plt.legend(['Actual', 'Predicted'])\n","plt.show()"],"metadata":{"id":"j2neqhK92Q7-"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["print(model.summary())"],"metadata":{"id":"L_7yZrEb3jzx"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["lstm_output = model.layers[0].output\n","dense_output = model.layers[2].output\n","\n","print(lstm_output)\n","print(dense_output)"],"metadata":{"id":"b2_BWsQ_4Kit"},"execution_count":null,"outputs":[]}]}