Title: Bitcoin Price Prediction using LSTM and Transformers
Description: This project focuses on scraping price-based data from various sources and building machine learning models for price predictions. The goal is to gather relevant textual information that can potentially impact the price of a particular asset, such as stocks, cryptocurrencies, or commodities, and use this data to train predictive models.
Key Steps:
1. Data Source Identification: Identify and select appropriate sources of text data that may contain relevant information for price predictions. These sources could include financial news websites, social media platforms, forums, or any other text-rich platforms.
2. Web Scraping: Develop web scraping scripts or tools to extract textual data from the identified sources. Use techniques like web crawling, API integration, or custom data extraction methods to collect the required information.
3. Data Preprocessing: Cleanse and preprocess the scraped text data to remove noise, handle missing values, perform normalization etc to prepare the data for further analysis.
4. Feature Extraction: Extract relevant features from the preprocessed data that can contribute to price predictions.
5. Machine Learning Modeling: Build machine learning models using the historical price data. Explore various algorithms such as regression, classification, or time series forecasting methods to develop accurate price prediction models.
6. Model Evaluation: Evaluate the performance of the developed machine learning models using appropriate evaluation metrics, cross-validation techniques, and backtesting on historical data. Fine-tune the models and iterate the process to improve the prediction accuracy.
7. Deployment and Monitoring: Deploy the trained models into a production environment where they can generate real-time price predictions. Implement monitoring mechanisms to track model performance, detect anomalies, and retrain/update the models as new data becomes available.
By leveraging text data scraping and machine learning techniques, this project aims to enhance price prediction capabilities by incorporating valuable textual information from various sources.
DICLAIMER: The resulting models does not in any way provide insights to decision-making processes for traders, investors, and financial analysts!
[INFO]: The tweets.parquet file is found here: https://drive.google.com/file/d/1tdtkIA3rR72lam0WwAGSgXefRcmgurtL/view?usp=drive_link
[INFO]: The tweets.csv file is found here: https://drive.google.com/file/d/1NY8_t4jiVczB_XuVEVGOxFbKiCyRRdGQ/view?usp=drive_link