LSTM stands for Long Short Term Memory Networks. It is a type of recurrent neural network that is commonly used for regression and time series forecasting in machine learning. It can memorize data for long periods, which differentiates LSTM neural networks from other neural networks. If you want to learn how to predict stock prices with LSTM, this article is for you. In this article, I will walk you through the task of stock price prediction with LSTM using Python.
Stock Price Prediction with LSTM
Using LSTM is one of the best machine learning approaches for time series forecasting. LSTMs are recurrent neural networks designed to remember data for a longer period. So, whenever you are working on a problem where your neural network fails to memorize data, you can use LSTM neural network. You can read more about LSTMs here.
Now in this section, I will take you through the task of Stock price prediction with LSTM using the Python programming language. I will start this task by importing all the necessary Python libraries and collecting the latest stock price data of Apple:
import pandas as pd
import yfinance as yf
import datetime
from datetime import date, timedelta
today = date.today()
d1 = today.strftime("%Y-%m-%d")
end_date = d1
d2 = date.today() - timedelta(days=5000)
d2 = d2.strftime("%Y-%m-%d")
start_date = d2
data = yf.download('AAPL',
start=start_date,
end=end_date,
progress=False)
data["Date"] = data.index
data = data[["Date", "Open", "High", "Low", "Close",
"Adj Close", "Volume"]]
data.reset_index(drop=True, inplace=True)
data.tail()Date Open High ... Close Adj Close Volume 3441 2021-12-27 177.089996 180.419998 ... 180.330002 180.330002 74919600 3442 2021-12-28 180.160004 181.330002 ... 179.289993 179.289993 79144300 3443 2021-12-29 179.330002 180.630005 ... 179.380005 179.380005 62348900 3444 2021-12-30 179.470001 180.570007 ... 178.199997 178.199997 59773000 3445 2021-12-31 178.089996 179.229996 ... 177.570007 177.570007 64025500 [5 rows x 7 columns]
A candlestick chart gives a clear picture of the increase and decrease in stock prices, so let’s visualize a candlestick chart of the data before moving further:
import plotly.graph_objects as go
figure = go.Figure(data=[go.Candlestick(x=data["Date"],
open=data["Open"],
high=data["High"],
low=data["Low"],
close=data["Close"])])
figure.update_layout(title = "Apple Stock Price Analysis",
xaxis_rangeslider_visible=False)
figure.show()
Now let’s have a look at the correlation of all the columns with the Close column as it is the target column:
correlation = data.corr() print(correlation["Close"].sort_values(ascending=False))
Close 1.000000 Low 0.999890 High 0.999887 Adj Close 0.999845 Open 0.999783 Volume -0.496325 Name: Close, dtype: float64
Training LSTM for Stock Price Prediction
Now I will start with training an LSTM model for predicting stock prices. I will first split the data into training and test sets:
x = data[["Open", "High", "Low", "Volume"]] y = data["Close"] x = x.to_numpy() y = y.to_numpy() y = y.reshape(-1, 1) from sklearn.model_selection import train_test_split xtrain, xtest, ytrain, ytest = train_test_split(x, y, test_size=0.2, random_state=42)
Now I will prepare a neural network architecture for LSTM:
from keras.models import Sequential from keras.layers import Dense, LSTM model = Sequential() model.add(LSTM(128, return_sequences=True, input_shape= (xtrain.shape[1], 1))) model.add(LSTM(64, return_sequences=False)) model.add(Dense(25)) model.add(Dense(1)) model.summary()
Model: "sequential_6"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
lstm_12 (LSTM) (None, 4, 128) 66560
lstm_13 (LSTM) (None, 64) 49408
dense_12 (Dense) (None, 25) 1625
dense_13 (Dense) (None, 1) 26
=================================================================
Total params: 117,619
Trainable params: 117,619
Non-trainable params: 0
_________________________________________________________________
Now here is how we can train our neural network model for stock price prediction:
model.compile(optimizer='adam', loss='mean_squared_error') model.fit(xtrain, ytrain, batch_size=1, epochs=30)
Epoch 1/30 2756/2756 [==============================] - 18s 5ms/step - loss: 6.7984 Epoch 2/30 2756/2756 [==============================] - 15s 5ms/step - loss: 4.4978 Epoch 3/30 2756/2756 [==============================] - 15s 5ms/step - loss: 5.6511 Epoch 4/30 2756/2756 [==============================] - 15s 5ms/step - loss: 6.8347 Epoch 5/30 2756/2756 [==============================] - 15s 5ms/step - loss: 9.5083 Epoch 6/30 2756/2756 [==============================] - 15s 5ms/step - loss: 7.4367 Epoch 7/30 2756/2756 [==============================] - 15s 5ms/step - loss: 4.3043 Epoch 8/30 2756/2756 [==============================] - 15s 6ms/step - loss: 4.2213 Epoch 9/30 2756/2756 [==============================] - 15s 5ms/step - loss: 5.7352 Epoch 10/30 2756/2756 [==============================] - 15s 6ms/step - loss: 5.2137 Epoch 11/30 2756/2756 [==============================] - 15s 5ms/step - loss: 6.0945 Epoch 12/30 2756/2756 [==============================] - 14s 5ms/step - loss: 4.1032 Epoch 13/30 2756/2756 [==============================] - 15s 5ms/step - loss: 4.3637 Epoch 14/30 2756/2756 [==============================] - 15s 5ms/step - loss: 6.2240 Epoch 15/30 2756/2756 [==============================] - 15s 5ms/step - loss: 1.9857 Epoch 16/30 2756/2756 [==============================] - 15s 5ms/step - loss: 6.3982 Epoch 17/30 2756/2756 [==============================] - 15s 5ms/step - loss: 3.3015 Epoch 18/30 2756/2756 [==============================] - 15s 5ms/step - loss: 3.9104 Epoch 19/30 2756/2756 [==============================] - 15s 5ms/step - loss: 4.6564 Epoch 20/30 2756/2756 [==============================] - 15s 5ms/step - loss: 3.3215 Epoch 21/30 2756/2756 [==============================] - 15s 6ms/step - loss: 4.3116 Epoch 22/30 2756/2756 [==============================] - 15s 5ms/step - loss: 2.8147 Epoch 23/30 2756/2756 [==============================] - 15s 5ms/step - loss: 5.7586 Epoch 24/30 2756/2756 [==============================] - 16s 6ms/step - loss: 4.1890 Epoch 25/30 2756/2756 [==============================] - 17s 6ms/step - loss: 3.6991 Epoch 26/30 2756/2756 [==============================] - 15s 6ms/step - loss: 4.0951 Epoch 27/30 2756/2756 [==============================] - 15s 5ms/step - loss: 3.5940 Epoch 28/30 2756/2756 [==============================] - 16s 6ms/step - loss: 3.7180 Epoch 29/30 2756/2756 [==============================] - 15s 6ms/step - loss: 3.5864 Epoch 30/30 2756/2756 [==============================] - 15s 5ms/step - loss: 3.7422 <keras.callbacks.History at 0x7f8c37686790>
Now let’s test this model by giving input values according to the features that we have used to train this model and predicting the final result:
import numpy as np #features = [Open, High, Low, Adj Close, Volume] features = np.array([[177.089996, 180.419998, 177.070007, 74919600]]) model.predict(features)
array([[179.95299]], dtype=float32)
So this is how we can use LSTM neural network architecture for the task of stock price prediction.
Summary
LSTM stands for Long Short Term Memory Networks. It is a recurrent neural network designed to remember data for longer. Using LSTM is one of the best machine learning approaches for time series forecasting. I hope you liked this article on predicting stock prices with LSTM using Python. Feel free to ask valuable questions in the comments section below.
