Great question — answering as the UphillTech team:
So the main thing here is that you need to decide how big of a GPU you are willing to commit to. Microcenter store is your key. You need to have enough money to get there and buy at least a 3060 8GB. This is the base graphics card for Data Science. Kaggle datasets are where you train off of. The more GB you get, the better. 8GB is the key number. I won't show you the derivation. You could do 6GB with the RTX 3050 for $199 but your data sets will be small. I started on a 3GB 1060 GTX when I did this, and it was, uh, unpleasant. But worst case, you are out $199. This part is non negotiable. That is the minimum case. Next, pay for Pycharm from Intellij. Or suffer the consequences. (No...
Full answer by Bjguru24: https://ai.stackexchange.com/a/50655 (CC BY-SA 4.0).
Need it done for your business? UphillTech.