Special Weekend session: Live Hands-On Lab model fitting with No-Code AI
Live lab walkthrough with AutoAI and Akk.io
Dear Class,
The AI Model theory class is posted as two one hour lesson on curriculum. Details now what topics were covered are in my post here. I have also listed the links to the recording below.
Theory Lesson : AI Model Part 1 is here (Feature engineering and Machine Learning Classification models)
Theory Lesson: AI Model Part 2 is here (Where business uses classification models, Neural Networks, Deep Learning and more)
We did a live lab walkthrough with a Telco Churn Dataset to build a binary classification model using IBM’s Auto AI and Akk.io as an experiment. (I have no affiliation and do not endorse any brands).
Dataset used for the lab: The dataset is here. Please download it and try it out for yourself.
Live Hands-on Lab: The live class lesson recording of the Live Lab walkthrough is here.
Here’s the model from IBM. You need to select one of the pipelines.
Here are steps we followed in the Live Hand-on AI Lab on IBM’s Auto.AI
Start with creating a free lite account on cloud.ibm.com
Add a Watson AI service. You need the one that says Machine Learning. There are quite a few Watson services for computer vision and for building a chatbot.
Create a project and attach the Watson AI service to it.
Add a data asset to the project and upload the telco churn dataset that we want to use in this experiment.
Start an Auto AI experiment, under your new project. Run it (with the Watson AI service and the data asset you added).
There are several sequences by which you can complete these steps. Don’t worry, you cannot break anything and need lots of patience.
The run can take unto 10 minutes on IBM AutoAI.
Now in parallel, goto Akk.io
Here are steps we followed in the Live Hand-on AI Lab on Akk.io
Start a new flow and give it a name
Input data and upload the same telco churn data file
Add a step and choose Predict
Run the Prediction
Here’s the model from akk.io
You should get an AI model from IBM AutoAI and Akk.io.
IBM gives you 10 pipelines to choose from to pick a model. Compare the confusion matrix for the accuracy of both models. Why do you think there is a difference between the two?
Time to wear your business hat
Next you can try your team’s industry data on each of these two platforms. First stop and think about what is the AI making a recommendation for and what can you do with it.
ok, if you were to deploy the model in your business, how would you use it.
Remember, a model is not a product. And how does it help your business?
Are there any decision points of trade-offs you need to take to use the models from either platform to work for your business? What do you need to do to make it solve your business problem that you set out to solve at the start of this Capstone AI Lab?
I would love to see your answers. You know where to find me.
Remember, I am here to help.