This project looked to forecast bitcoin prices in the very near future based on a wide variety of data points, from US economic indicator data to the stock price of major GPU manufacturers. I thought that their exploratory analysis alone was extremely interesting and the use of its results to develop a model exploration strategy thoughtful. Even when highlighting the relative inaccuracy of their results, their final discussion section was strong and raised interesting questions about the implications of openly available bitcoin prediction tools. Overall, the report is beautiful, very well written and contains several informative and professional visualizations.
I feel that the number of features used was a little low and would have liked to have seen a wider variety of feature types beyond just continuous. I felt that the correlation analysis which identified a small subset of features with a very high correlation to Bitcoin price could have been used as a tool to find these new features. As a side note, I would be interested to know whether a volatile commodity like Bitcoin had any similarities to a developed, more cyclical market like the S&P 500. Finally, although I liked the thorough discussion at the end, it would have been interesting to hear more potential users and use cases for your model.
Ultimately, the presentation of the report was incredible and the thought process used to develop each subsequent model extremely well described. Great work.
This project looked to forecast bitcoin prices in the very near future based on a wide variety of data points, from US economic indicator data to the stock price of major GPU manufacturers. I thought that their exploratory analysis alone was extremely interesting and the use of its results to develop a model exploration strategy thoughtful. Even when highlighting the relative inaccuracy of their results, their final discussion section was strong and raised interesting questions about the implications of openly available bitcoin prediction tools. Overall, the report is beautiful, very well written and contains several informative and professional visualizations.
I feel that the number of features used was a little low and would have liked to have seen a wider variety of feature types beyond just continuous. I felt that the correlation analysis which identified a small subset of features with a very high correlation to Bitcoin price could have been used as a tool to find these new features. As a side note, I would be interested to know whether a volatile commodity like Bitcoin had any similarities to a developed, more cyclical market like the S&P 500. Finally, although I liked the thorough discussion at the end, it would have been interesting to hear more potential users and use cases for your model.
Ultimately, the presentation of the report was incredible and the thought process used to develop each subsequent model extremely well described. Great work.