|
virtual const char * | GetClassName () |
|
virtual int | IsA (const char *type) |
|
virtual void | PrintSelf (ostream &os, vtkIndent indent) |
|
int | GetNumberOfAttributeArrays () |
| Return the number of columns available for the current value of AttributeMode. More...
|
|
const char * | GetAttributeArrayName (int n) |
| Get the name of the n-th array ffor the current value of AttributeMode. More...
|
|
int | GetAttributeArrayStatus (const char *arrName) |
| Get the status of the specified array (i.e., whether or not it is a column of interest). More...
|
|
vtkInformationIntegerKey * | MULTIPLE_MODELS () |
| A key used to mark the output model data object (output port 0) when it is a multiblock of models (any of which may be multiblock dataset themselves) as opposed to a multiblock dataset containing a single model. More...
|
|
|
virtual int | GetAttributeMode () |
| Set/get the type of field attribute (cell, point, field) More...
|
|
virtual void | SetAttributeMode (int) |
| Set/get the type of field attribute (cell, point, field) More...
|
|
|
void | EnableAttributeArray (const char *arrName) |
| An alternate interface for preparing a selection of arrays in ParaView. More...
|
|
void | ClearAttributeArrays () |
| An alternate interface for preparing a selection of arrays in ParaView. More...
|
|
|
virtual void | SetTrainingFraction (double) |
| Set/get the amount of data to be used for training. More...
|
|
virtual double | GetTrainingFraction () |
| Set/get the amount of data to be used for training. More...
|
|
|
virtual void | SetTask (int) |
| Set/get whether this filter should create a model of the input or assess the input or both. More...
|
|
virtual int | GetTask () |
| Set/get whether this filter should create a model of the input or assess the input or both. More...
|
|
|
| vtkSciVizStatistics () |
|
virtual | ~vtkSciVizStatistics () |
|
virtual int | FillInputPortInformation (int port, vtkInformation *info) |
|
virtual int | FillOutputPortInformation (int port, vtkInformation *info) |
|
virtual int | ProcessRequest (vtkInformation *request, vtkInformationVector **input, vtkInformationVector *output) |
|
virtual int | RequestDataObject (vtkInformation *request, vtkInformationVector **input, vtkInformationVector *output) |
|
virtual int | RequestData (vtkInformation *request, vtkInformationVector **input, vtkInformationVector *output) |
|
virtual int | RequestData (vtkCompositeDataSet *compDataOu, vtkCompositeDataSet *compModelOu, vtkCompositeDataSet *compDataIn, vtkCompositeDataSet *compModelIn, vtkDataObject *singleModel) |
|
virtual int | RequestData (vtkDataObject *observationsOut, vtkDataObject *modelOut, vtkDataObject *observationsIn, vtkDataObject *modelIn) |
|
virtual int | PrepareFullDataTable (vtkTable *table, vtkFieldData *dataAttrIn) |
|
virtual int | PrepareTrainingTable (vtkTable *trainingTable, vtkTable *fullDataTable, vtkIdType numObservations) |
|
virtual int | LearnAndDerive (vtkMultiBlockDataSet *model, vtkTable *inData)=0 |
| Method subclasses must override to calculate a full model from the given input data. More...
|
|
virtual int | AssessData (vtkTable *observations, vtkDataObject *dataset, vtkMultiBlockDataSet *model)=0 |
| Method subclasses must override to assess an input table given a model of the proper type. More...
|
|
virtual vtkIdType | GetNumberOfObservationsForTraining (vtkTable *observations) |
| Subclasses may (but need not) override this function to guarantee that some minimum number of observations are included in the training data. More...
|
|
Abstract base class for computing statistics with vtkStatistics.
This filter either computes a statistical model of a dataset or takes such a model as its second input. Then, the model (however it is obtained) may optionally be used to assess the input dataset.
This class serves as a base class that handles table conversion, interfacing with the array selection in the ParaView user interface, and provides a simplified interface to vtkStatisticsAlgorithm.
- Thanks:
- Thanks to David Thompson and Philippe Pebay from Sandia National Laboratories for implementing this class. Updated by Philippe Pebay, Kitware SAS 2012
Definition at line 51 of file vtkSciVizStatistics.h.
Possible tasks the filter can perform.
The MODEL_AND_ASSESS task is not recommended; you should never evaluate data with a model if that data was used to create the model. Doing so can result in a too-liberal estimate of model error, especially if overfitting occurs. Because we expect that MODEL_AND_ASSESS, despite being ill-advised, will be frequently used the TrainingFraction parameter has been created.
Enumerator |
---|
MODEL_INPUT | Execute Learn and Derive operations of a statistical engine on the input dataset.
|
CREATE_MODEL | Create a statistical model from a random subset the input dataset.
|
ASSESS_INPUT | Assess the input dataset using a statistical model from input port 1.
|
MODEL_AND_ASSESS | Create a statistical model of the input dataset and use it to assess the dataset.
This is a bad idea.
|
Definition at line 116 of file vtkSciVizStatistics.h.
virtual void vtkSciVizStatistics::SetTrainingFraction |
( |
double |
| ) |
|
|
virtual |
Set/get the amount of data to be used for training.
When 0.0 < TrainingFraction < 1.0, a randomly-sampled subset of the data is used for training. When an assessment is requested, all data (including the training data) is assessed, regardless of the value of TrainingFraction. The default value is 0.1.
The random sample of the original dataset (say, of size N) is obtained by choosing N random numbers in [0,1). Any sample where the random number is less than TrainingFraction is included in the training data. Samples are then randomly added or removed from the training data until it is the desired size.
virtual double vtkSciVizStatistics::GetTrainingFraction |
( |
| ) |
|
|
virtual |
Set/get the amount of data to be used for training.
When 0.0 < TrainingFraction < 1.0, a randomly-sampled subset of the data is used for training. When an assessment is requested, all data (including the training data) is assessed, regardless of the value of TrainingFraction. The default value is 0.1.
The random sample of the original dataset (say, of size N) is obtained by choosing N random numbers in [0,1). Any sample where the random number is less than TrainingFraction is included in the training data. Samples are then randomly added or removed from the training data until it is the desired size.
virtual int vtkSciVizStatistics::AssessData |
( |
vtkTable * |
observations, |
|
|
vtkDataObject * |
dataset, |
|
|
vtkMultiBlockDataSet * |
model |
|
) |
| |
|
protectedpure virtual |
Method subclasses must override to assess an input table given a model of the proper type.
The dataset parameter contains a shallow copy of input port 0 and should be modified to include the assessment.
Adding new arrays to point/cell/vertex/edge data should not pose a problem, but any alterations to the dataset itself will probably require that you create a deep copy before modification.
- Parameters
-
observations | - a table containing the field data of the dataset converted to a table |
dataset | - a shallow copy of the input dataset that should be altered to include an assessment of the output. |
model | - the statistical model with which to assess the observations. |
Implemented in vtkPSciVizMultiCorrelativeStats, vtkPSciVizKMeans, vtkPSciVizPCAStats, vtkPSciVizContingencyStats, and vtkPSciVizDescriptiveStats.
virtual vtkIdType vtkSciVizStatistics::GetNumberOfObservationsForTraining |
( |
vtkTable * |
observations | ) |
|
|
protectedvirtual |
Subclasses may (but need not) override this function to guarantee that some minimum number of observations are included in the training data.
By default, it returns the maximum of: observations->GetNumberOfRows() * this->TrainingFraction and min( observations->GetNumberOfRows(), 100 ). Thus, it will require the entire set of observations unless there are more than 100.
[in] observations - a table containing the full number of available observations (in this process).