VIDEO | Getting Started with ValidMind QuickStart

This short video walks users through documenting a model with ValidMind. This demo shows how to run our QuickStart notebook on JuypyterHub, and show you how to work with model documentation on the ValidMind platform.

QuickStart Transcript:

From docs.validmind.com go to QuickStart right above my head, and we recommend that you try the JupyterHub QuickStart as it contains everything that you need. If you don’t already have an account, you’re going to need to create one to follow along. Hop over to JupyterHub and what you get is a notebook that contains both code and explanations for what the code does. Before you can do anything, you need to install the client library by running the cell and we will also need a model to document so let’s hop over into our platform UI, the Model Inventory, and register a new model. We’re going to call that QuickStart Demo. Binary classification is the template, and the use case is Churn Management.

Now, once this model is registered in the inventory we already get some bare bones documentation, but the sections are currently empty. You can see here, just placeholders — we’re going to fix that. To fix this, we need our code snippet to connect ValidMind and the notebook. So, this the code snippet. I’ve copied it. I’m going to go and paste this into the notebook, and then I’m going to restart the kernel and run all cells in this notebook. And what this does is it loads a sample data set to train a simple classification model and then it runs a ValidMind test suite to quickly generate documentation about the data and the model based on the template that we picked. You can see here we’re connected to ValidMind — that’s good. We’re previewing the template with different sections including what we saw on the platform UI before. That’s where it’s loading the data set, and I’m just going to scroll down. This might take anywhere between 30 seconds and a minute to run, and I’m going to rejoin you in a moment here once the full test suite has finished running.

Alright, so that test suite has now finished running. You can see some of the output here. You can look at this output right in the notebook, but a much better way is to go back into the platform UI and to go back into the documentation we looked at earlier, and now you can see for example that the data description section is filled out with some of the tests that we ran. You can also see that some tests are satisfactory if I scroll down you can see here that some tests like here the Skewness test will require more attention. And that, in a nutshell, is how you document models with ValidMind. If you like what you see, try the QuickStart.