The official implementation of the paper "CryptoPulse: Short-Term Cryptocurrency Forecasting with Dual-Prediction and Cross-Correlated Market Indicators". Portions of the code have been adapted from the DLinear implementation.

Below is the list of the top 20 cryptocurrencies (by market capitalization at the time of data collection), along with the exact CSV file names (which include the -USD ticker format). The numeric prefix in each CSV file indicates its rank. You can see these file names in the dataset folder:

BTC-USD (1.BTC-USD.csv)

ETH-USD (2.ETH-USD.csv)

USDT-USD (3.USDT-USD.csv)

BNB-USD (4.BNB-USD.csv)

SOL-USD (5.SOL-USD.csv)

STETH-USD (6.STETH-USD.csv)

XRP-USD (7.XRP-USD.csv)

USDC-USD (8.USDC-USD.csv)

DOGE-USD (9.DOGE-USD.csv)

TON11419-USD (10.TON11419-USD.csv)

ADA-USD (11.ADA-USD.csv)

SHIB-USD (12.SHIB-USD.csv)

AVAX-USD (13.AVAX-USD.csv)

WSTETH-USD (14.WSTETH-USD.csv)

WETH-USD (15.WETH-USD.csv)

DOT-USD (16.DOT-USD.csv)

LINK-USD (17.LINK-USD.csv)

WBTC-USD (18.WBTC-USD.csv)

TRX-USD (19.TRX-USD.csv)

WTRX-USD (20.WTRX-USD.csv)

• Short-Term Forecasting: Predict cryptocurrency prices with a model designed for short-term predictions(next-day).
• Batch Processing: Run predictions for multiple cryptocurrencies at once.
• Flexible & Easy to Use: Install, run, and watch the results roll in!

You can install the package directly from GitHub with pip:

    pip install git+https://github.com/aamitssharma07/SAL-Cryptopulse.git

Alternatively, you can install it manually:

1. Clone the repo:

    git clone https://github.com/aamitssharma07/SAL-Cryptopulse.git
   
    cd SAL-Cryptopulse
   

2. Install dependencies in a virtual environment:

    python -m venv env_crypto
   
    env_crypto\Scripts\activate
   
    pip install -e .
   

After installation, you can run the model directly using the package's command line interface if you had installed the package directly from GitHub with `pip :

For a single run:

    
    cryptopulse --data BTC-USD --exp-name cryptopulse_results --ob-len 7 --pred-len 1 --batch-size 8 --learning-rate 0.0005 --train-epochs 100

For batch processing (multiple cryptos at once):

    cryptopulse_batch

Alternatively, you can run the model using the Python command if you have cloned the repo in your local

For a single run:

    python -m cryptopulse.main --data BTC-USD --exp-name cryptopulse_results --ob-len 7 --pred-len 1 --batch-size 8 --learning-rate 0.0005 --train-epochs 100

For batch processing (multiple cryptos at once):

    python -m cryptopulse.batch_processor

All results are saved in the results/cryptopulse_results directory.

Feel free to use this research work. Let’s build a smarter CryptoPulse together! You can connect with me on LinkedIn: https://www.linkedin.com/in/aamit-datascientist/

@INPROCEEDINGS{10982029, author={Kumar, Amit and Ji, Taoran}, booktitle={2024 IEEE International Conference on Big Data (BigData)}, title={CryptoPulse: Short-Term Cryptocurrency Forecasting with Dual-Prediction and Cross-Correlated Market Indicators}, year={2024}, volume={}, number={}, pages={1-8}, keywords={}, doi={10.1109/BigData62323.2024.10982029}}