I’ve followed the instructions here to set up my miniconda ml4t environment - https://github.com/PacktPublishing/Machine-Learning-for-Algorithmic-Trading-Second-Edition/tree/master/installation/windows… None of the notebooks are working when I launch jupyter notebook after using cd to get to the location… Am I missing something… I found this page incredibly confusing… https://github.com/PacktPublishing/Machine-Learning-for-Algorithmic-Trading-Second-Edition/tree/master/installation… Do we still need to use this docker thing? I just want to run the notebooks and play around with the data…
1 Like
I’m going to be honest I was in the same boat as you. None of these really work turnkey. You’ll find yourself getting frustrated and beating your head against the wall. Get a firm understanding of Python and re-write/modify them as necessary in Colab would be my advice.
1 Like
Thx Jeremy, I just don’t get why there can’t be a step by step on how to get the notebooks to work on a common operating system like windows. I actually bought the book a year ago and gave up reading it as it was impossible to follow. Recently stumbled across this youtube review Machine Learning for Algorithmic Trading by Stefan Jansen - Book Review - YouTube which stated that you can easily get the notebooks to work (which I strongly disagree with). The other frustrating thing about the book is the lack of sample data to get a sense of what format it needs to be in. I have been coding in python for 5+ years and have access to a Bloomberg terminal so I don’t need all of the emphasis in the book on scraping free data.I just want to be able to play around with working notebooks which is what I thought I was buying when I purchased the book. I’ve used anaconda navigator to launch the ml4t env and a lot of the code works but there are a lot of ImportError: No module named ‘talib’ etc… I thought the idea of downloading the env was so that all of the libraries were available in the env??
Wow this isn’t encouraging. I am having a tough time myself. Perhaps it’s because the libraries have been updated since, etc. I couldn’t install talib from my conda environment either, and couldn’t find it. To add insult to injury, my installation of seaborn wasn’t even recognized by the notebooks, for whatever reason. I saw something about hidden code, and using the yaml file to set up the virtual environment is cool (slick really), but also removes a bit of what is going on from your consciousness in my mind. I’m using Interactive Brokers and their Python API which I have learned a good amount on through Mayank Rasu’s Udemy class, which was amazing. Very detailed and walked you through it all from start to finish really. But, no real ML going on which is the point of a project I am working on and need to fold into that code. I will keep forging ahead of course, any advice/feedback/snickers in the hallway greatly appreciated
My experience with coding and python basically started when I bought the kindle version of this book last year. I use a Mac and more recently google colab, so unfortunately can’t say much from a Windows perspective. I literally had no idea what a virtual environment was when I started out, but with enough google searches and whatnot, I was able to get the notebooks running. But yeah, nothing feels very turnkey, but once you get things running there are only a few minor hiccups here and there. But, I also can run it in google colab with a mounted google drive, so that might be an easier option.
2 Likes
I agree with @craigwesley figure out how to get it running in Colab with a mounted drive. I will say if you cannot figure out how to get Seaborn working you have a long path ahead of you. I do not mean this as a dig but this is a rather easy Google search fix. Getting these notebooks running in Colab will not be the easiest thing to do but well worth the time.
1 Like
Thanks for your response Jeremy. I am interested in what you refer to as an easy Google search fix. I know how to use Seaborn and have been using it in Jupyter Lab, IDLE, etc. for some time, it appears that for whatever reason it’s just not making it from Conda to the Jupyter instance. Can you provide a hint then at what I should be searching for? Of course more than happy/willing to do the work and I realize sometimes the struggle is necessary, but maybe I’m missing something? If you have some knowledge I’m all ears. Thank you.
Jeff
1 Like
pip install __________
Or in Colab
!pip install __________
Thanks for your response Jeremy. Unfortunately that wasn’t the question/issue… I shared some screenshots where you could see seaborn was installed and functioning. I’m thinking the environment being used in Jupyter wasn’t that where it was installed. I did do some more digging afterwards and suspect that to be the issue but I’m using Spyder now for whatever that’s worth. Definitely worth figuring out after I get done with my current project/class and do appreciate your input/suggestion regarding Colab, thank you! I do like the book and expect that rewriting the code will help with my gaps in Python knowledge for sure.
Jeff
1 Like
I bookmarked the ML4T github years ago. Recently reignite my hope to pick it up. Tried to install the env in both Linux (Windows wsl) and Windows env. The installations went smooth. But seems nothing works.
- Jupyter notebook: can not open any notebook. Some local host error (http://localhost:8080). Do not know why
- Spyder: the env to install Spyder is not ready yet. Some Qt5… issue.
Do not know what the installations are about. And do not know how to move forward.