Summary

In this chapter, we've built our first multiclass classification model. After a brief introduction to the use case, we discovered what multiclass logistic regression is and how it can be used to classify events, behaviors, and objects according to their features into more than two categories.

Before diving into the development of the ML model, we analyzed the schema of the dataset related to the trees in New York City and applied some data quality checks necessary to build an effective ML model.

During the training stage, we trained three different ML models using different features to gradually improve the performance of the BigQuery ML model.

Then, we chose the third ML model and we evaluated it against the evaluation dataset. In this phase, we noticed that the ML model was able to maintain its performance on new records also and was ready to pass to the next phase.

In the last step, we used our ML model to classify the trees in New York City into five different...