CryptoTREND prediction

Inspiration

Inspired by the need to analyze and predict trends in Cryptocurrency, for Hackalytics 2023, we decided to apply machine learning and data analysis techniques to predict the trend of Bitcoin's market value.

What it does

For this project, we wrote a Jupyter Notebook that preprocessed dataset of historical cryptocurrency prices, fitted the data into a LSTM based model, attempted to predict the trend of Bitcoin in the 2013-2022 timeline and evaluated the model's precision.

How we built it

Before the project, we had little to no prior knowledge in data analysis and machine learning. Therefore, this project was an experimental learning experience for us to learn more about various data preprocessing and analysis techniques as well as useful concept such as LSTM and time series analysis.

Challenges we ran into

As we did more research on the topic, we realized that, due to the highly unpredictable and unstable of cryptocurrency, a LSTM based model might not be suitable for making accurately cryptocurrency trend forecast as it relied on sequence prediction rather than inherently randomness. We tried to produce the results regardless and was satisfied what we have learned along the journey.

Accomplishments that we're proud of

In a 36-hour timespan, we were able to do research on data preprocessing, time series analysis and LSTM model fitting. We were satisfied with what we have learned and realized the potential and exciting world of machine learning and data analysis.

What's next for CryptoTREND prediction

This project has provided us the background knowledge to expand to many exciting projects, such as text recognition, weather forecasting or image/video analysis. We gained a better understanding of data analysis and machine learning and can't wait to apply our knowledge to future projects.

Built With

• kaggle
• lstm
• python