transportation

BigQuery ML Fare Prediction

Fare Prediction Without the ML Infrastructure

BigQuery ML Fare Prediction

Fare Prediction Without the ML Infrastructure

SQL-Only

ML Pipeline

Zero

External Infrastructure

Production

Grade Accuracy

The Challenge

Ride-hailing companies want to tell passengers "your trip will cost $X" before the ride starts. Normally that requires a whole ML infrastructure — training servers, model serving, feature pipelines. We wanted to see if we could skip all that and do it in SQL.

Our Approach

Turns out, you can. We trained a regression model entirely inside BigQuery using just SQL queries. No separate ML infrastructure, no model serving pipeline. The model lives where the data lives. We validated with RMSE/MAE and built Python dashboards to visualize predictions.

Tech Stack

BigQueryBigQuery MLPythonGCP