This is a complementary repository for a manuscript in review. We include our dataset alongside the scripts that where used for its creation. This study was done for the Bitcoin, Ethereum, Litecoin and Ripple cryptocurrencies. However you can run the scripts yourself to collect data from any desired cryptocurrency. Machine Learning models where implemented using the Sci-kit library.
-
Python 3+
-
InfluxDB
-
Redis
Recomended
- Jupyter Notebook
You can play with the provided datasets. If you want to collect your own data you can use the fetch_market_data.py script and Tweet-catcher.
Datasets are included under the data folder in .h5 format.
Processing scripts are contained inside the scripts folder.
-
fetch_market_data.py --s1 LTC --s2 USD --days 80- Gets data from now since the specified number of days for the symbol pair. -
fetcher.py --days 60- Creates a queue for obtaining tweets from now since the specified number of days. You need to run a worker instance. -
process_tweets.py- Applies VADER lexicon to the tweets -
process_data.py- Concatenates the market and twitter data into a single dataset