Difference between revisions of "Measures"

From SUMOwiki
Line 23: Line 23:
|name = storeModels
|name = randomThreshold
|values = boolean
|values = positive integer
|default = no
|default = 1000
|description = When this option is turned on, the models built for each fold are stored on disk for future reference.
|description = If the number of samples is greater than this number a random partitioning is used

Revision as of 10:51, 17 September 2008

A measure is used to measure the quality of a new model. The measure decides whether the modeling process can be halted (the target accuracy has been reached) or whether new samples should be selected / new models should be built. The choice of measure is therefore very important for the success of the toolbox.

The rule of thumb is that the default measure, CrossValidation, is the best choice, but is also very expensive, as it requires that a number (5 by default) of models are built for each new model that has to be evaluated. Especially when using neural networks this can become an unacceptable overhead. If the modeling takes unacceptably long, the best alternative is ValidationSet, which by default behaves as cross validation in which only one fold is considered, reducing the cost of the measure by a factor 5. A second measure, called MinMax, is also activated by default, enabling the user to force the model to remain within certain bounds, to speed up the convergence. See below for more details in how to set these bounds.

However, in certain situations it might be very effective to use different measures, or use multiple measures together. When multiple measures are used, an intelligent pareto-based method is used to decide which model is the best choice. Models that score high on a particular measure but low on another are not discarded immediately, but are given a chance to set things right in further iterations of the toolbox. This encourages variety in the models, while still ensuring convergence to the optimal accuracy for each measure. An often used combination is CrossValidation with the MinMax measure, to ensure that no poles are present in the model domain.

Note that also an error function can be defined. The default error function is 'rootRelativeSquareError'.

Below is a list of available measures and the configuration options available for each of them. Each measure also has a target accuracy attribute, which can be omitted and which defaults to 0.001. In certain cases, such as the binary MinMax measure, the target accuracy is irrelevant.


The CrossValidation measure is the default choice and performs an n-fold cross validation on the model to create an efficient estimation of the accuracy of the model. Several options are available to customize this measure.

option values default
folds positive integer 5
The number of folds used for the measure. A higher number means that more models will be built, but that a better accuracy estimate is achieved.
randomThreshold positive integer 1000
If the number of samples is greater than this number a random partitioning is used
partitionMethod [uniform,random] uniform
This option defines whether the test sets for the folds are chosen randomly, or are chosen in such a way as to maximize the domain coverage. Random is generally much faster, but might result in pessimistic scoring, as unlucky test set choice can result in an inaccurate error. This can partly be fixed by enabling the resetFolds option.
resetFolds boolean no
Folds are generated from scratch for each model that is evaluated using this measure. If the same model is evaluated twice (for example, after a rebuild), new folds are used. Enabling this feature can be very costly for large sample sizes. As a rule of thumb, enable this in combination with the random partition method, or disable this when using the uniform method.


The ValidationSet measure has two different methods of operation.

  1. In the first method, the list of samples that have been evaluated is split into a validation set and a training set. A model is then built using the training set, and evaluated using the validation set (which is by default 20% of the total sample pool).
  2. However, an external data file containing a validation set can also specified. In this case, all the evaluated samples are used for training, and the external set is used for validation only. Which of these two operation methods is used, depends on the configuration options below. By default, no external validation set is loaded.

If you want to use an external validation set, you will have to provide a SampleEvaluator configuration so that the validation set can be loaded from an external source. Here is a ValidationSet configuration example which loads the validation set from the scattered data file provided in the simulator file:

<Measure type="ValidationSet" target=".001">
	<Option key="type" value="file"/>
	<SampleEvaluator type="ibbt.sumo.SampleEvaluators.ScatteredDatasetSampleEvaluator"/>
option values default
type [distance, random, file] distance
Method used to acquire samples for the validation set. The default method, 'distance', tries to select a validation set which covers the entire domain as good as possible, ensuring that not all validation samples are chosen in the same part of the domain. This is achieved using a distance heuristic, which gives no guarantees on optimal coverage but performs very well in almost all situations. The 'random' method just picks a random set of samples from the entire pool to be used for validation set. Finally, the 'file' method does not take samples at all from the pool, but loads a validation set from an external dataset.
percentUsed [0,100] 20
Percent of samples used for the validation set. By default 20% of all samples are used for validation, while the remaining 80% are used for training. This option is irrelevant if the 'type' option is set to 'file'.
randomThreshold positive integer 1000
When the sample pool is very large, the distance heuristic used by default becomes too slow, and the toolbox switches to random sample selection automatically. This is done when the amount of samples is larger than this value, which defaults to 1000. This option should not be changed unless the performance is unacceptable even for sample sets smaller than this amount.



The MinMax measure is used to eliminate models whose response falls below a given minimum or above a given maximum. This measure can be used to detect models that have poles in the model domain and to guide the modeling process in the right direction. If the output is known to lie within certain value bounds, these can be added to the simulator file as follows:

	<Parameter name="out" type="real" minimum="-1" maximum="1"/>

When the MinMax measure is defined, these values will be used to ensure that all models stay within these bounds. If only the minimum or only the maximum is defined, naturally only these are enforced. There are no further configuration options for this measure. In case of complex outputs the modulus is used.

Remember though, that no guarantee can be given that the poles will really disappear. Using this measure only combats the symptoms and not the cause of the problem. Also, this measure can be reasonably slow, because it evaluates a dense grid to decide wether the model is crossing boundaries. If the model is slow to evaluate, this can take a considerable amount of time.

Tip: even if you don't know the exact bounds on your output, you can still use this measure by specifying very broad bounds (e.g., [-10000 10000]). This can still allow you to catch poles since, by definition, they reach until infinity.