Your Path to AI: Fraud Prediction using AutoAI

Your Path to AI: Fraud Prediction using AutoAI

Your Path to AI is a series of workshops to get you started on your path to becoming proficient in AI. Each session consists of important topics in the Data & AI methodology so that you can learn relevant concepts and understand what tools IBM provides along with open-source to make Your Path to AI easier.

Now wouldn’t it be convenient and stress-free to watch your Machine Learning pipeline workloads to be completely automated? With all this growth in the field of data science and AI, what if I tell you IBM’s offering, AutoAI can easily perform automated data preparation, apply ML algorithms, and build model pipelines best suited for your datasets.

On 18th October 2020, in the fourth session of Your Path to AI series, IBM Developer Advocates Anam Mahmood and Sidra Ahmed talk about how you can easily build Machine Learning models using AutoAI and integrate them with IBM CloudPak for Data.

The workshop was divided into two parts where the first part was conducted by Sidra Ahmed, she kicked it off by explaining the foundational concepts behind Artificial Intelligence, Machine Learning, and Deep Learning using a timeline. And further went into explaining the different Machine Learning algorithms.

Machine Learning Algorithms

She then introduced AutoML, why it’s important, it’s benefits, and how it makes a difference as compared to traditional Machine Learning.

Benefits of Automated Machine Learning

Sidra then gave an introduction to AutoAI, an IBM Cloud offering, that makes building machine learning pipelines easy. She explained the different stages of the Data Science pipeline that AutoAI automates for us and the benefits of using AutoAI.

Benefits of using AutoAI

The second part of the workshop was conducted by Anam Mahmood where she demonstrated how to use AutoAI to build a machine learning model to predict the insurance premium charges. She started by familiarizing the attendees with IBM Cloud and then created the Watson Studio service, Cloud Object Storage, and Watson Machine Learning service. She then went ahead and explained the architecture of what she would be building and for this session they focused on the cloud element.

Architecture of the workflow

The flow description:

  1. The user creates an IBM Watson Studio Service on IBM Cloud.
  2. The user creates an IBM Cloud Object Storage Service and adds that to Watson Studio.
  3. The user uploads the insurance premium data file into Watson Studio.
  4. The user creates an AutoAI Experiment to predict insurance premium on Watson Studio
  5. AutoAI uses Watson Machine Learning to create several models, and the user deploys the best performing model.
  6. The user uses the Flask web-application to connect to the deployed model and predict an insurance charge.

Anam then created an AutoAI experiment, gave the audience a tour of the AutoAI tool showing them the entire process from uploading the dataset to choosing the best pipeline based on different parameters to model deployment. She further explained how different factors such as smoking and BMI affect the model prediction and you will see a radical difference the lifestyle choices can make on insurance charges.

Testing the deployment of the model

Using IBM AutoAI, we automate all the tasks involved in building predictive models for different requirements. AutoAI generates great models quickly, which saves time and effort, and aids in a much faster decision-making process.

The attendees followed along with the code pattern. Anam and Sidra answered all their queries. The workshop received a great amount of positive feedback and they enjoyed learning an easy way to generate models.


Cloud Developer Advocate at IBM