Skip to content

Latest commit

 

History

History
41 lines (27 loc) · 2.72 KB

README.md

File metadata and controls

41 lines (27 loc) · 2.72 KB
Raw

Stock Price Prediction Using LSTM

This project demonstrates the application of Long Short-Term Memory (LSTM) networks for predicting stock prices, specifically for Open, High, Low, and Volume of stock data. It uses a univariate and multivariate approach to make predictions based on historical stock data.

Project Overview

This project uses a dataset from Amazon's stock price data (Amazon Data.csv). The dataset contains stock information including the Open, High, Low, Close, and Volume for each day. The goal is to train a model to predict future stock prices and volumes using LSTM networks, a type of Recurrent Neural Network (RNN) that is effective for time series forecasting.

Data Cleaning

  • The dataset is cleaned by setting the Date column as the index and converting it into a datetime object.
  • Missing values are handled by applying a rolling mean with a window of 5 to fill gaps in data for the Open, Close, and Volume columns.
  • After cleaning, the dataset is prepared for training and visualization.

Univariate Stock Price Prediction

The project first uses a univariate approach, where each feature (Open, High, Low, and Volume) is predicted using only its own past values. The data is scaled using MinMaxScaler, then split into training and testing datasets. The model is built using LSTM layers, trained for 10 epochs with a batch size of 1.

Functions:

  1. uni_variate_training(target) - Trains a univariate LSTM model on the given target column (e.g., 'Open', 'High').
  2. validate_and_plot(target, model, ylabel) - Validates the model by predicting on test data and visualizing the results.

Multivariate Stock Price Prediction

The project also performs multivariate analysis, where the model takes into account multiple features (Open, High, Low, Close, Volume) to predict the Close price. The data is similarly scaled and split, but the input to the LSTM model is multidimensional (5 features over 45 days).

Function:

  • multivariate_training() - Trains a multivariate LSTM model using multiple features to predict stock prices.

Interactive Visualizations

The project includes an interactive web application using Streamlit that allows users to:

  • View the cleaned dataset.
  • View statistical descriptions of the dataset.
  • Select and test different models (univariate and multivariate).
  • Visualize the predicted values against actual values for various stock features.

Features:

  • Univariate Prediction Models: Predicts stock features individually.
  • Multivariate Prediction Model: Predicts the closing stock price using multiple features.
  • Visualization: Displays training and testing data with predicted stock values.