Tensorflow.js tf.metrics.meanAbsolutePercentageError() Function

Tensorflow.js is an open-source library developed by Google for running machine learning models as well as deep learning neural networks in the browser or node environment.

The .metrics.meanAbsolutePercentageError() function is a loss or else a metric function i.e. mean absolute percentage error which uses truth and prediction tensor inputs in order to return tf.Tensor object.

Syntax:  

tf.metrics.meanAbsolutePercentageError(yTrue, yPred)

Parameters:  

• yTrue: It is the stated truth tensor and it can be of type tf.Tensor.
• yPred: It is the stated prediction tensor and it can be of type tf.Tensor.

Return Value: It returns the tf.Tensor object.

Example 1:  

Javascript

// Importing the tensorflow.js library
import * as tf from "@tensorflow/tfjs"

// Defining truth and prediction tensors
const y = tf.tensor2d([[0, 2], [20, 30]]);
const z = tf.tensor2d([[0, 2], [21, 34]]);

// Calling metrics.meanAbsolutePercentageError()
// method
const mape = tf.metrics.meanAbsolutePercentageError(y, z);

// Printing output
mape.print();

Output:

Tensor
    [0, 9.166666]

Example 2:

Javascript

// Importing the tensorflow.js library
import * as tf from "@tensorflow/tfjs"

// Calling metrics.meanAbsolutePercentageError()
// method with its parameter directly and then
// Printing output
const output = tf.metrics.meanAbsolutePercentageError(tf.tensor(
    [
      [0, 1, 0, 0],
      [0, 1, 1, 0],
      [0, 0, 0, 1],
      [1, 1, 0, 0],
      [0, 0, 1, 0]
    ]
), tf.tensor(
    [
      [0, 0, 1, 1],
      [0, 1, 1, 0],
      [0, 0, 0, 1],
      [0, 1, 0, 1],
      [1, 1, 0, 0]
    ]
)).print();

Output:

Tensor
    [500025, 0, 0, 250025, 500025]

Reference: https://js.tensorflow.org/api/latest/#metrics.meanAbsolutePercentageError