Chapter 3: Linear regression ... stations names?

PacktPublishing / Machine-Learning-with-BigQuery-ML

Machine Learning with BigQuery ML, published by Packt

MIT License

30 stars 20 forks source link

Chapter 3: Linear regression ... stations names? #3

Closed kolban-google closed 2 years ago

kolban-google commented 2 years ago

I have a question. I read chapter 4 on linear regression using the CitiBikes story. I'm at a loss to understand how using the string place names of the start and end locations could ever be used as features for linear regression? I don't think there is an underlying set of equations that could be used to predict duration given two textual names.

sathyarams commented 2 years ago

@kolban-google https://cloud.google.com/bigquery-ml/docs/reference/standard-sql/bigqueryml-auto-preprocessing should provide an explanation. For linear regression, all non-numeric values will be one-hot encoded.