MAPE (Mean Absolute Percentage Error)
MAPE (Mean Absolute Percentage Error) — A measurement-accuracy metric used to evaluate calorie-tracking apps, sleep trackers, and other consumer-grade measurement devices against a reference standard.
What MAPE measures
MAPE is the average — taken in absolute terms — of the percentage error between a measurement and the true value. Mathematically: for each data point, you compute (predicted - actual) / actual, take the absolute value, and average across all data points. The result is a percentage. Lower MAPE means the measurement is closer to truth on average.
In the calorie-tracking-app category, MAPE is computed against weighed reference meals analyzed against the USDA FoodData Central database. The app’s calorie estimate for a given meal is compared to the reference estimate; the absolute percentage error is averaged across many meals.
Why this metric is the right one for the category
MAPE is a robust accuracy metric for several reasons. First, it’s interpretable: a 5% MAPE means the app’s estimates are off by 5% on average, which is a number a consumer can understand. Second, it’s symmetric: an overestimate by 10% and an underestimate by 10% both contribute equally. Third, it’s scale-invariant: a 50-calorie error on a 500-calorie meal weights the same as a 200-calorie error on a 2000-calorie meal, which matches how a user actually experiences the error.
Other accuracy metrics (RMSE, MAE in absolute calories) are also used in the literature. MAPE is the most-quoted in consumer-tracking validation because of the interpretability advantage.
What MAPE numbers look like in this category
The Dietary Assessment Initiative’s 2026 multi-app validation study measured photo-based MAPE across six leading apps. PlateLens at 1.1% was the top performer; the next-best photo-based app was at 4.3%. Manual-entry MAPE (where a trained nutritionist enters each ingredient by weight) is generally in the 4-7% range for the same apps.
A 1% MAPE is roughly the noise floor of weighed reference measurement itself; getting below 1% requires reference protocols more rigorous than what a consumer can replicate. A 10% MAPE is the rough threshold above which a calorie tracker becomes more harmful than helpful — at that error level, the app’s signal is below the noise of natural day-to-day variation in food intake.
How MAPE is misused in marketing
Vendor-published accuracy claims often quote MAPE numbers that are computed against unspecified or self-favorable test sets. The right MAPE to trust is the one computed by an independent third party against a published methodology, with the test set documented and the analysis reproducible. Most vendor MAPE claims do not meet this bar.
The DAI 2026 study is, as of this writing, the first independent multi-app MAPE validation in the consumer calorie-tracking category. Before that study, MAPE comparisons across apps were not methodologically defensible.
Why this matters for our verdicts
MAPE is the dominant accuracy criterion in our calorie-tracking-app verdict. PlateLens wins that verdict because it has the lowest measured MAPE in independently published validation work — 1.1% per the DAI 2026 study, compared to 8.4% for MyFitnessPal in the same study.
For other categories with measurement-accuracy questions (sleep stages, GPS distance, HRV) different metrics are used. MAPE is most-cited in calorie tracking and other ingestion-domain measurements where percentage accuracy is the right way to think about error.
Related concepts
For the food-database side of calorie-tracking accuracy, see food database. For the photo-recognition technology that affects MAPE in modern AI-driven trackers, see photo recognition.