Mean Absolute Percentage Error (MAPE) Metric¶

The mape metric calculates the Mean Absolute Percentage Error, measuring the accuracy of predicted values against observed values as a percentage. MAPE provides a scale-independent measure of prediction accuracy that is easy to interpret.

Metric Type: mape

MAPE Calculation¶

The MAPE is calculated as: mean(|observed - predicted| / |observed|) * 100

Where:

observed = Observed values (actual values)
predicted = Predicted values
The result is expressed as a percentage

MAPE is a relative measure that expresses accuracy as a percentage, making it useful for comparing models across different scales. A value of 0% indicates perfect prediction accuracy.

Configuration Fields¶

Record-Level Data Format¶

For individual observation records:

metrics:
  model_mape:
    metric_type: "mape"
    config:
      name: ["prediction_accuracy"]
      data_format: "record_level"
      observed: "observed_values" # Column with observed/actual values
      predicted: "predicted_values" # Column with predicted values
      segment: [["model_version"]] # Optional: segmentation columns
      dataset: "predictions"

Summary-Level Data Format¶

For pre-aggregated percentage error data:

metrics:
  summary_mape:
    metric_type: "mape"
    config:
      name: ["aggregated_mape"]
      data_format: "summary_level"
      volume: "observation_count" # Column with observation counts
      sum_absolute_percentage_errors: "sape" # Column with sum of absolute percentage errors
      segment: [["data_source"]] # Optional: segmentation columns
      dataset: "error_summary"

Required Fields by Format¶

Record-Level Required¶

name: Metric name(s)
data_format: Must be "record_level"
observed: Observed values column name
predicted: Predicted values column name
dataset: Dataset reference

Summary-Level Required¶

name: Metric name(s)
data_format: Must be "summary_level"
volume: Volume count column name
sum_absolute_percentage_errors: Sum of absolute percentage errors column name
dataset: Dataset reference

Optional Fields¶

segment: List of column names for grouping

Output Columns¶

The metric produces the following output columns:

group_key: Segmentation group identifier (struct of segment values)
volume: Total number of observations
mape: Mean Absolute Percentage Error value (as percentage)

Fan-out Examples¶

Single Configuration¶

metrics:
  basic_mape:
    metric_type: "mape"
    config:
      name: ["model_mape"]
      data_format: "record_level"
      observed: "actual_values"
      predicted: "predicted_values"
      dataset: "validation_data"

Segmented Analysis¶

metrics:
  segmented_mape:
    metric_type: "mape"
    config:
      name: ["regional_mape", "product_mape"]
      data_format: "record_level"
      observed: "observed_values"
      predicted: "predicted_values"
      segment: [["region"], ["product_type"]]
      dataset: "performance_data"

Mixed Data Formats¶

metrics:
  detailed_mape:
    metric_type: "mape"
    config:
      name: ["record_level_mape"]
      data_format: "record_level"
      observed: "actual"
      predicted: "predicted"
      dataset: "detailed_data"

  summary_mape:
    metric_type: "mape"
    config:
      name: ["summary_mape"]
      data_format: "summary_level"
      volume: "count"
      sum_absolute_percentage_errors: "sape"
      dataset: "summary_data"

Data Requirements¶

Record-Level Data¶

One row per observation
Observed column: numeric values (must be non-zero to avoid division by zero)
Predicted column: numeric values (any numeric value is allowed)
Observed values should not contain zeros or values very close to zero

Summary-Level Data¶

One row per group/segment
Volume counts: positive numbers
Sum of absolute percentage errors: positive numbers

MAPE Interpretation¶

Value Guidelines¶

0%: Perfect prediction accuracy
< 10%: Highly accurate predictions
10-20%: Good prediction accuracy
20-50%: Reasonable prediction accuracy
> 50%: Poor prediction accuracy

Scale Independence¶

MAPE is scale-independent, making it useful for comparing predictions across different data ranges
Results are expressed as percentages, making them intuitive to interpret
Can be used to compare model performance across different datasets

Important Notes¶

Zero Division: MAPE cannot be calculated when observed values are zero or very close to zero
Asymmetric: MAPE is asymmetric - over-predictions and under-predictions are treated differently
Scale Independence: Unlike RMSE, MAPE is not affected by the scale of the data
Percentage Format: Results are expressed as percentages (0-100+)
Data Quality: Remove observations where observed values are zero or missing before calculation
Outlier Sensitivity: MAPE can be sensitive to outliers, especially when observed values are small