Fixing Garmin’s chaotic data exports: New tool preps smartwatch metrics for custom analytics

For many smartwatch buyers, Garmin's ecosystem could be a strong selling point. It provides versatile analytics covering both individual workouts and overall training status. While the platform isn't entirely open, exporting session data remains relatively straightforward.

Analyzing that exported data independently, however, is another story. It typically requires building a custom analytical pipeline or feeding the raw files into Large Language Models. The export files present several technical hurdles: incomplete fields, missing measurements, and inconsistent data types can easily disrupt computer-assisted parsing.

This is where the Garmin Running Data Normalizer comes in. Available as a PyPI package, the tool cleans and standardizes raw export data from Garmin wearables. While it doesn't analyze the data itself, it transforms chaotic exports into structured datasets ready for downstream evaluation.

Processing is executed entirely locally, keeping sensitive fitness metrics and running routines on the user's machine. Some limitations remain—for instance, Health Status data is not supported in the current release. Comprehensive documentation is available, though using the tool requires comfort with the command line.