Data Science Setup
Okay, debatable. You can go with VS Code too. But just check this out.
Local Runtime Google Colab setup
Get the best of both worlds¶
Why set up a local instance when you have google providing you powerful CPUs in the cloud that you can use for training highly complex models that require a lot of resources?
Simply because it is slow for the analysis part. Most of the time is spent in data cleaning than actually running the model. So, I will say that there are two modes of working:
- Data preparation and Exploratory Analysis
- Working with models
For (1) it is better to use a local instance as it gives you
- quick autocompletes
- ability to run "light" code snippets quickly (like doing data frame operations)
- offline access to your code
For (2) it is better to use a cloud instance as it gives you
- power
- freedom to do something else on your device while the model runs
Local Instance Setup¶
For using Tensorflow 2 in machine learning projects, we can download miniconda to get python installed and then load up this environment. This can be activated from anywhere and be used for all datascience projects. A really good base which includes most of the things needed