I set up this blog to help me out on my machine learning journey.
I have already lots of experience with Python and data wrangling with Pandas and other libraries, but now I want to get back to machine learning and see what I can do.
A few years ago I learned the basics of ML with fast.ai, a library and online course that really make things easy to get started.
Fast.ai
Fast.ai underneath use pytorch for the heavy lifting but also implement the most commons algorithms and utilited needed to buid a complete working prototype. Things like pretrained model and training data downloading and caching are all builtin and work without lifting a finger. Usefull algorithms that optimise training speed and accuracy, data splitting for training, tests and validation.
This time I want to continue with Fast.ai, but also go a bit deeper into Pytorch and take a look at TensorFlow also. I read that TensorFlow used to be much harder to use but that has changed now and the new version are closer to how Pytorch work.
GamestonkTerminal
Another interesting project is the Gamestonk Terminal. This is a open source trading terminal inspired by the Bloomberg terminal. This project gained a lot of traction as the Bloomberg terminal is very expensive and thus out of reach for most people. The Gamestonk Terminal gained a lot of contributions since it’s inception and in addition of the traditional technical analysis now contains machine learning prediction models and algo testing system.I want to try it out and see if I make use of it for trading and for machine learning.
Machine lerning setup
I will take a look at the differents way to set up my environment for machine learning. For now I tried Colab form Google, it has been underwhelming soo far. I tried it a while ago, and I found out that it is still not standart to have your own disk tied to your running terminal, even with the pro version. I will thus take a look again at doing my own setup with Ubuntu and a NVidia card and see the other for machine learning on the cloud.
Good luck with your experiments and see you next time.
