Auto ML: Making your first AutoML model

Welcome back to Obviously AI university. 

If you are watching these videos sequentially you’ll have recently watched the videos on how to format your data so it is machine learning ready.

We are now going to focus on building a highly accurate classification Auto ML model.  

From the top-down, we are going to cover building your Machine learning model, predicting outcomes, understanding factors that affect model performance, making predictions via batch and API, evaluating model performance, understanding the technical specifications of the model and updating and sharing your model.

After watching these videos you’ll have an understanding of how to integrate AI and Machine learning into an overall business strategy. 

We’re using the same dataset as in our “Adding Data” videos which is a loan repayment dataset. 

We’ve already gone in and cleaned everything up, because we know from our previous videos how important it is to have machine learning-ready datasets. 

To quickly review this dataset, here it is, you’ll see it contains all sorts of information, such as the customer’s age, id, income, loan intent, loan grade, and more. 

This column here, Status, is the prediction column - it shows whether the loan was repaid or not. 

Now, let’s start building our Auto ML model. 

  • After uploading your dataset, you’re taken to the review page where we see an overview of the datasets. We see the column is fit for predictions, so, let’s Click continue, where we are taken to a screen to pick out prediction column.
  • Since we’re trying to determine whether someone will default on their payment or not, we’re going to select status to predict. As we can see, the platform indicates that this column is indeed fit for prediction. 
  • We now click “Start Predicting” and wait for the magic to happen. 

 Even though it only takes a matter of seconds, there’s a lot going on behind the scenes here. During this time, Obviously AI is running the dataset through final pre-processing stages and is training multiple models.  The backend will run various hyperparameter combinations of top performing algorithms. This essentially means trying multiple algorithms with different settings, mixing and matching them resulting in 10000+ of algorithms running parallely, to find the one that performs the best. 

The platform then shortlists the best algorithms that are fit to be used on your dataset. 

And just like that, you're taken to the reports page, and are now ready to make predictions!

That’s how fast and easy it is. We have a model trained and we’ll soon be generating prediction reports in a fraction of the time it would take if built using traditional code. 

For further videos and educational content subscribe to our YouTube channel today to check out our comprehensive Obviously AI University course and learn more about how you can leverage no-code AI to transform your business. If you'd like to see it in action, click the demo link in the description below"

Complete and Continue