JupyterLab — nearly as good as Colab

JupyterLab I’ve noticed has had quite a few features added recently. A few months ago it was quite buggy, GUI was average and weren’t able to install certain package versions (e.g I wasn’t able to install the latest pandas profiling).

Colab was quite the opposite and was way ahead of JupyterLab. I have done a small write up on Colab here

Looks like Google have released some features to JupyterLab and have addressed some of the issues I previously had.

JupyterLab is starting to slowly look like Colab’s user-friendly interface which is great for users.

Colab does not cost (paid version for more compute is available), but JupyterLab does cost no matter what (depends on the size of the compute chosen). A basic low cloud compute option costs roughly US$99.89 monthly, $0.137 hourly (for Machine type: 4 vCPUs, 15 GB RAM, Boot disk: 100 GB Disk). You can turn the instance off and on, so you only pay for the time you are executing code.

Ease of mind using private data

JupyterLab is set up within your own Private Google cloud, permissions to read your private data are sorted and the security risks are significantly lowered. The only thing to mention is that it costs.

Hopefully one my articles helps someone! Complexity is the enemy to execution. Born and bred kiwi