Eduard Samokhvalov

- curves of pnl - some risk metrics of the portfolio like VaR and the ones we already have for single model tests - some portfolio optimization library being easily...

> if you mean some kind of clustering, then CatBoost is not applicable directly to build a model, CatBoost needs a training dataset which contains enough examples for each equivalence...

Oh thanks, I see. How about 'indirectly', can it propose some type of classification that would make sense?

Thank you, Evgueni.

Guys, it still won't work. Could you help with a step by step manual? Conda is not an option for me. PS C:\Users\Administrator\Desktop\seasonalities\ftp> pip install prophet Collecting prophet Using cached...

@akosfurton guys, you are the best! it finally worked :) May I also ask what would be the best way to run seasonalities without producing output in the console, thus...

Thank you!

@Evgueni-Petrov-aka-espetrov Thanks for helping out. So most categorical features have around 10 values. Those: "TOP_Departure", "Downtrend_Starting", "Downtrend_Neutral", "Downtrend_ApproachingBottom", "BOTTOM_Arrival","Uptrend_Starting", "Uptrend_Neutral", "Uptrend_ApproachingTop", "BOTTOM_Departure", "TOP_Arrival" (the phases of cycle). Then the target...

Great, here's the data https://filetransfer.io/data-package/b920vRWq#link Then I run the code below like this python .\catboost_from_tutorial.py .\crypto_cycle5_long_7days.csv front_3 (trying to forecast column front_3) ```python import pandas as pd import os import...

So the error I get in the run (took around 2 days) is somewhat around 0.68 or 0.56, pretty high error