Supporting a New Dataset

Extending the pipeline to new projects

So far, we’ve built an ML training pipeline and trained it on one specific problem—iris classification. A pipeline that works for a single project is good, but a pipeline that can extend to multiple projects is significantly more useful because it saves development effort.

We’ve used our pipeline to train a model to predict the species of irises. Can we extend it to new projects, such as predicting fuel consumption for automobiles using the well-known AutoMPG dataset? We can! The following diagram shows a graphical view of how this is achieved.