How to Manage Conda Environments for Data Science

Key Takeaways

managing environment is crucial in your role as a data scientist and in this video I will show you a few essential commands that you need to know in order to work with environments now a conda environment is a directory that contains a specific collection of packages that you have installed first make sure that Anaconda is installed properly by opening up a terminal or command prompt and typing kunda info if this works you're good to go otherwise go and install Anaconda first to create an environment we'll run conda create dash n followed by the name of the environment and then we can also specify a certain python version if everything looks good we can press Y and hit enter this will set up the environment and install the dependencies and we can now see our new demo environment here to activate our new alignment we run conda activate followed by the environment name the environment name between parentheses before your username indicates that the environment is active we can now run conda list to see what is inside of the environment and as you can see there are only a handful of packages installed as a bare minimum to make this environment useful we have to install some packages you can either install packages using pip or kunda and if you want to learn more about the differences then please check out this article which I will link in the description I will now install pandas and matplotlip using Pip this will install the latest versions of both packages into the environment we can now run conda list again to validate the installations you can see that we have pandas a mottled lip installed as well as some other dependencies that are required for multiple blip or bundles now if we actually want to use this environment we can open up an IDE like Visual Studio code for example and here I'm inside in a project and all the way down in the corner over here you can see which environment we are using and what python version we're running but what we can do if we click on this we can select another environment from the list of environments and as you can see our demo environment that we just created is now listed here and if we select this environment we can now run this project using this environment so let me switch back to the environment that we were using so that was the bundles tutorials one because I will now show you how to export this environment and this is as easy as running the command called.nf export let me show you so I open up a terminal we're in the bundles tutorials environment and then we type conda and Export then this Arrow sign and make sure to add the dot yml extension now hit enter and you can see that we have a new environment file over here now if I open this up you can see that everything is in here so we have the name of the environment we have the channels all the dependencies so this is a complete export of your environment with all the dependencies packages and version numbers now we can use this file to create a completely new environment with a single line of code and that's why it's very important to keep this environment.yml file up to date especially prior to installing major new packages that could potentially break your environment so what happened to me this week I was working on a machine learning project for a client and past week I've made some good progress but what happened so at 10 o'clock so right before the meeting I was experimenting with a different kind of model so a new method that could potentially also benefit the project and I try to install this package in the environment using pip and what happened the whole environment broke I couldn't even fire up a Jupiter interactive session anymore and now I set that with a broken environment and a meeting in about an hour where I had to present my results so let me show you how this works I first deactivate this environment now one more thing that I have to do is to change the name of the environment because this one already exists so I go to the file and instead of bonus tutorials I create bundles tutorials too make sure to save this then go back now run the command conda and create Dash F then link to the file it will collect all the package metadata this could take some time because it's installing a lot of things once it's done we can run Konda and list to validate and we can see that we have a new environment bundles tutorials too in the list now let's select it within vs code so go to The Interpreter and then select bundles tutorials 2. now this simple Act of exporting your environment prior to making any major changes to the environment could have saved me a big headache so for your next projects get into the habit of regularly exporting your environment files you will thank yourself later and you can also easily share these environments with your colleagues working on the same project so now you know how to manage combat environments and why it's important to keep them up to date thanks for watching and if you want to learn more about data science then subscribe to the channel and check out some of the other videos that will pop up on the screen now