I followed my planned workout fully, but the system has classified it incorrectly. Why?

Understanding exactly what type of run you've done is a sophisticated task for any system. The immense diversity in how people run, the wide variety of workouts, and the variable quality of data from consumer GPS devices all contribute to making run type classification a significant (mathematical) challenge. Even for our team at TrainAsONE, some runs are very tricky to categorise, and we acknowledge that the system (and us!) can sometimes make mistakes.

This type of problem is known in AI as 'pattern recognition,' similar to the complexities of teaching computers to read different handwriting styles or identify objects in images. At present, our classification accuracy is a little over 80%, though this isn't uniform. Some individuals experience much better accuracy, while for others, it's lower. So, an occasional misclassification is to be expected, and nothing to be concerned about.

Your input is crucial here. When the system does get it wrong, please use the run confirmation widget to correct it. This feedback is vital for our AI, as it learns and improves from more high-quality data. We have ongoing algorithm enhancements planned, and we're optimistic about achieving over 95% accuracy for all users in the not-too-distant future.

Importantly, the classification of your run currently serves only as a minor safety check within the plan generation process; it's not a major driver of your future workouts (in fact its effect rapidly diminishes after 24 hours). Our ultimate aim is for run classification to become purely an informational tool for your reference, with no direct influence on your training plan - we are nearly at that point.

If you need to correct a run, our 'My run is classified incorrectly. How do I correct it? FAQ has detailed steps.