AutoAI is a graphical tool for automatically prepares data, applies algorithms, or estimators, and builds model pipelines that are best suited for your data and use case.
IBM introduce many strategies for automated model development, which displays the results as model candidate pipelines ranked on a leaderboard. It could serve non-technical users such as domain experts and business users to build and deploy machine learning models. AutoAI also implements automated feature engineering and hyperparameter optimization.
AutoAI has success cases lending in many industries. In healthcare, AutoAI has been used in heart failure rate prediction and feature importance for sudden deafness. During this session, we will share a worldwide use case of using AutoAI for blood donation prediction. In addition, a real dataset will be introduced to showcase how AutoAI quickly develop models. Researchers only need to customize the business goals (in this case, the blood donation time), AutoAI will then builds models to satisfy the goal while optimizing for the model performance (e.g., ROC AUC score). It is evident from the results, that with a minimum span of time, the application is automatically modeled and deployed for the use case.
Lastly, IBM Trustworthy AI toolkit, Watson Openscale, will be incorporated to provide the explainability of the model. Using the fairness, quality and drift indicators, the researchers can easily verify that the AI model meets the desired standard.