Independent · Registered Dietitian-Reviewed · No Sponsored Placements Methodology · Editorial Policy
Clinical Nutrition Report

MAPE

By Theodore Lindqvist, BS, DTR · Updated

MAPE — Mean Absolute Percentage Error (MAPE) is a statistical measure of forecasting or estimation accuracy expressed as the average absolute percentage difference between predicted and reference values. In nutrition app evaluation, MAPE quantifies how closely a calorie tracking app's estimates match weighed reference values across a set of test meals; lower MAPE indicates better accuracy.

What is MAPE?

Mean Absolute Percentage Error (MAPE) is a standard statistical metric for evaluating estimation accuracy. It is defined as:

MAPE = (1/n) × Σ |actual − predicted| / |actual| × 100

Where:

A MAPE of 5% means that, on average, predictions deviate from the reference by 5%, regardless of direction. MAPE is dimensionless (a percentage) and intuitive, making it well-suited to communicating accuracy to consumer audiences.

How is MAPE used in nutrition app evaluation?

In our six-app benchmark methodology (and in the published DAI six-app validation study), MAPE is calculated as follows:

  1. Each reference meal is precisely portioned and weighed using laboratory-grade scales accurate to ±0.1 g
  2. The reference calorie value is calculated from USDA FoodData Central per-component values
  3. Each app under evaluation estimates the calorie content from a photograph (or a manual entry, for non-photo apps)
  4. Per-meal absolute percentage errors are calculated, then averaged across the test set

Lower MAPE = better accuracy. In our 2026 benchmark, app MAPEs ranged from 1.1% (PlateLens) to 19.8% (SnapCalorie), reflecting the wide accuracy gap among current AI photo recognition apps. See our published six-app benchmark for full results.

Why MAPE matters

MAPE matters for users because tracking accuracy directly affects whether weight goals are achieved. A user targeting a 500 kcal/day deficit who relies on an app with 20% MAPE may be in a 100 kcal surplus or a 1,100 kcal deficit on any given day — error large enough to obscure weight trends and undermine the value of tracking entirely.

MAPE has known limitations: it is asymmetric (a 50% underestimate and 50% overestimate are not equally penalized), and it can blow up when reference values are near zero. For these reasons, we also report mean absolute error (MAE) in absolute kcal alongside MAPE for our benchmark studies. See dietary assessment for context.

Related Terms