Bitcoin Price Prediction

Inspiration

The major inspiration for the project was the current talk about Bitcoin in the media. We are all interested in the concept of Cryptocurrency and how we think that cryptocurrency is the future.

What it does

This project takes the past recorded history of Bitcoin and attempts to model the fluctuation in pricing. We use the past histories and trends within the history to try and predict whether or not the price will rise or fall over the course of the next hour. We also tried to use machine learning through the implementation of a Linear SVM representation to also model the future Bitcoin prices.

How I built it

Through using Python, we were able to create and test functionality of our algorithm to test whether or not the price would increase. We also used Python to create a webscraper using BeautifulSoup to parse websites for the past history of Bitcoin and saved this information into a .csv file. Using the .csv file to train our data, we were able to create a prediction that fairly represents the future pricing of Bitcoin. We used tkinter to create a GUI that we could use to make this information more presentable.

Challenges I ran into

We had issues with the Linear SVM modeling when it came to the machine learning portion of the project and we also had some issues with creating the GUI.

Accomplishments that I'm proud of

That we learned how to use a webscraper and we have learned significantly more about machine learning through this process.

What I learned

The ins and outs with machine learning and how difficult it is to create and accurately model our information with very few data points. Also, we learned just how unpredictable Bitcoin is.

What's next for Bitcoin Price Prediction

A better GUI that can predict past values based upon all of the information given to the program before the given time frame. We also want to better estimate our prices based upon the machine learning aspect of this project. We could also add functionality to examine more data points to create better training data for both the machine learning algorithm and our own implementation of our algorithm.