« blib/lib/Paws/SageMaker/AutoMLJobConfig.pm
↑ Summary
blib/lib/Paws/SageMaker/AutoMLJobSummary.pm »
| 1 | # Generated by default/object.tt | ||
| 2 | package Paws::SageMaker::AutoMLJobObjective; | ||
| 3 | use Moose; | ||
| 4 | has MetricName => (is => 'ro', isa => 'Str', required => 1); | ||
| 5 | |||
| 6 | 1; | ||
| 7 | |||
| 8 | ### main pod documentation begin ### | ||
| 9 | |||
| 10 | =head1 NAME | ||
| 11 | |||
| 12 | Paws::SageMaker::AutoMLJobObjective | ||
| 13 | |||
| 14 | =head1 USAGE | ||
| 15 | |||
| 16 | This class represents one of two things: | ||
| 17 | |||
| 18 | =head3 Arguments in a call to a service | ||
| 19 | |||
| 20 | Use the attributes of this class as arguments to methods. You shouldn't make instances of this class. | ||
| 21 | Each attribute should be used as a named argument in the calls that expect this type of object. | ||
| 22 | |||
| 23 | As an example, if Att1 is expected to be a Paws::SageMaker::AutoMLJobObjective object: | ||
| 24 | |||
| 25 | $service_obj->Method(Att1 => { MetricName => $value, ..., MetricName => $value }); | ||
| 26 | |||
| 27 | =head3 Results returned from an API call | ||
| 28 | |||
| 29 | Use accessors for each attribute. If Att1 is expected to be an Paws::SageMaker::AutoMLJobObjective object: | ||
| 30 | |||
| 31 | $result = $service_obj->Method(...); | ||
| 32 | $result->Att1->MetricName | ||
| 33 | |||
| 34 | =head1 DESCRIPTION | ||
| 35 | |||
| 36 | Specifies a metric to minimize or maximize as the objective of a job. | ||
| 37 | |||
| 38 | =head1 ATTRIBUTES | ||
| 39 | |||
| 40 | |||
| 41 | =head2 B<REQUIRED> MetricName => Str | ||
| 42 | |||
| 43 | The name of the objective metric used to measure the predictive quality | ||
| 44 | of a machine learning system. This metric is optimized during training | ||
| 45 | to provide the best estimate for model parameter values from data. | ||
| 46 | |||
| 47 | Here are the options: | ||
| 48 | |||
| 49 | =over | ||
| 50 | |||
| 51 | =item * | ||
| 52 | |||
| 53 | C<MSE>: The mean squared error (MSE) is the average of the squared | ||
| 54 | differences between the predicted and actual values. It is used for | ||
| 55 | regression. MSE values are always positive: the better a model is at | ||
| 56 | predicting the actual values, the smaller the MSE value. When the data | ||
| 57 | contains outliers, they tend to dominate the MSE, which might cause | ||
| 58 | subpar prediction performance. | ||
| 59 | |||
| 60 | =item * | ||
| 61 | |||
| 62 | C<Accuracy>: The ratio of the number of correctly classified items to | ||
| 63 | the total number of (correctly and incorrectly) classified items. It is | ||
| 64 | used for binary and multiclass classification. It measures how close | ||
| 65 | the predicted class values are to the actual values. Accuracy values | ||
| 66 | vary between zero and one: one indicates perfect accuracy and zero | ||
| 67 | indicates perfect inaccuracy. | ||
| 68 | |||
| 69 | =item * | ||
| 70 | |||
| 71 | C<F1>: The F1 score is the harmonic mean of the precision and recall. | ||
| 72 | It is used for binary classification into classes traditionally | ||
| 73 | referred to as positive and negative. Predictions are said to be true | ||
| 74 | when they match their actual (correct) class and false when they do | ||
| 75 | not. Precision is the ratio of the true positive predictions to all | ||
| 76 | positive predictions (including the false positives) in a data set and | ||
| 77 | measures the quality of the prediction when it predicts the positive | ||
| 78 | class. Recall (or sensitivity) is the ratio of the true positive | ||
| 79 | predictions to all actual positive instances and measures how | ||
| 80 | completely a model predicts the actual class members in a data set. The | ||
| 81 | standard F1 score weighs precision and recall equally. But which metric | ||
| 82 | is paramount typically depends on specific aspects of a problem. F1 | ||
| 83 | scores vary between zero and one: one indicates the best possible | ||
| 84 | performance and zero the worst. | ||
| 85 | |||
| 86 | =item * | ||
| 87 | |||
| 88 | C<AUC>: The area under the curve (AUC) metric is used to compare and | ||
| 89 | evaluate binary classification by algorithms such as logistic | ||
| 90 | regression that return probabilities. A threshold is needed to map the | ||
| 91 | probabilities into classifications. The relevant curve is the receiver | ||
| 92 | operating characteristic curve that plots the true positive rate (TPR) | ||
| 93 | of predictions (or recall) against the false positive rate (FPR) as a | ||
| 94 | function of the threshold value, above which a prediction is considered | ||
| 95 | positive. Increasing the threshold results in fewer false positives but | ||
| 96 | more false negatives. AUC is the area under this receiver operating | ||
| 97 | characteristic curve and so provides an aggregated measure of the model | ||
| 98 | performance across all possible classification thresholds. The AUC | ||
| 99 | score can also be interpreted as the probability that a randomly | ||
| 100 | selected positive data point is more likely to be predicted positive | ||
| 101 | than a randomly selected negative example. AUC scores vary between zero | ||
| 102 | and one: a score of one indicates perfect accuracy and a score of one | ||
| 103 | half indicates that the prediction is not better than a random | ||
| 104 | classifier. Values under one half predict less accurately than a random | ||
| 105 | predictor. But such consistently bad predictors can simply be inverted | ||
| 106 | to obtain better than random predictors. | ||
| 107 | |||
| 108 | =item * | ||
| 109 | |||
| 110 | C<F1macro>: The F1macro score applies F1 scoring to multiclass | ||
| 111 | classification. In this context, you have multiple classes to predict. | ||
| 112 | You just calculate the precision and recall for each class as you did | ||
| 113 | for the positive class in binary classification. Then, use these values | ||
| 114 | to calculate the F1 score for each class and average them to obtain the | ||
| 115 | F1macro score. F1macro scores vary between zero and one: one indicates | ||
| 116 | the best possible performance and zero the worst. | ||
| 117 | |||
| 118 | =back | ||
| 119 | |||
| 120 | If you do not specify a metric explicitly, the default behavior is to | ||
| 121 | automatically use: | ||
| 122 | |||
| 123 | =over | ||
| 124 | |||
| 125 | =item * | ||
| 126 | |||
| 127 | C<MSE>: for regression. | ||
| 128 | |||
| 129 | =item * | ||
| 130 | |||
| 131 | C<F1>: for binary classification | ||
| 132 | |||
| 133 | =item * | ||
| 134 | |||
| 135 | C<Accuracy>: for multiclass classification. | ||
| 136 | |||
| 137 | =back | ||
| 138 | |||
| 139 | |||
| 140 | |||
| 141 | |||
| 142 | =head1 SEE ALSO | ||
| 143 | |||
| 144 | This class forms part of L<Paws>, describing an object used in L<Paws::SageMaker> | ||
| 145 | |||
| 146 | =head1 BUGS and CONTRIBUTIONS | ||
| 147 | |||
| 148 | The source code is located here: L<https://github.com/pplu/aws-sdk-perl> | ||
| 149 | |||
| 150 | Please report bugs to: L<https://github.com/pplu/aws-sdk-perl/issues> | ||
| 151 | |||
| 152 | =cut | ||
| 153 |