2
votes
The Folder where you can locate the data & the code (using Jupyter).
https://notebooks.azure.com/febelaz/libraries/akademi4en-dataton
The Folder where you can locate the data & the code (using Jupyter).
https://notebooks.azure.com/febelaz/libraries/akademi4en-dataton
3 thoughts on “THE A.I. CRYPTO TRADER: cryptomonkeys”
Seems like you have worked methodically. It would have been much better if you published the final mathematical form of the best model. Also why did you only forecasted bitcoin? How do you know that this is the most important crypto?
I really like that you have approached the data prep correctly.
As it seems you have put a lot of effort in model building so as to deliver a sound predictive model at the end of the day. Therefore, congrets!
Yet, I would like to add up several comments that might help you in your future work as data analysts. First of all, it is always advised to accompany every presented result with a relevant comment. Thus the reader might clearly and easily follow the whole process.
Secondly, it is always a good idea to motivate the core research steps conducted, e.g. what motivates the choice of methodology adopted; what are the competing scenarios, etc.
Overall, good job, guys!
Great job. The article looks very good. One thing to fix – make it all in English.
For the data clean and prep you did good job – I could recommend that you may do some of the following checks: outliers, periodic, trend.
It is great that you implemented ARIMA model. Optional you could try SWR, PCA or GARCH.
Another important part of the task is the metrics.
Keep up with the job and I am sure other mentors will give you more comments and recommendations.