perClass Documentation
development version 3.1.2 (22-Dec-2011)
Content

Comments? Ideas? Compliments?

Your email (only if you wish to be contacted)

perClass Matlab Toolbox Interface

Table of contents

1. Examples and help ↩

2. sddata class: Handling data sets ↩

  • sddata - Data set object
  • sdimport - Import sddata object from text file
  • sdexport - Export sddata object into text file
  • getdata - Get data matrix
  • double - Convert to double
  • getname - Get data set name string
  • setname - Set data set name string
  • size - Get data set size (number of samples, features and classes)
  • dataset - Convert sddata object into PRTools dataset

2.1. Data subsets ↩

  • subset - Subset of samples given label/property values
  • find - Get indices of samples by label/property values
  • randsubset - Random subset sampling classes or any property

2.2. Interactive data visualization ↩

  • sdscatter - Interactive scatter plot; hand-painting class labels; filtering samples
  • sdimage - Image plot with hand-painting of image labels
  • sdfeatplot - Interactive plot of per-feature class distributions

2.3. Handling data set labels ↩

  • getlab - Get current labels
  • setlab - Set a property as labels or find what property is used as labels
  • getpriors - Get vector of prior probabilities for current labels
  • getsizes - Get vector of class sizes for current labels
  • isclass - Test if class or classes are present
  • sdrelab - Rename classes or define meta-classes

2.4. Handling data set feature labels ↩

  • getfeatlab - Get current feature labels
  • setfeatlab - Set a property as feature labels or find what property is used as labels

2.5. Handling data set properties ↩

  • getprop - Get property
  • setprop - Set (create) property
  • rmprop - Remove property from data set
  • getproplist - Get list of properties available in data set
  • isprop - Test if property is present

2.6. Operations on data sets ↩

  • abs - Absolute value
  • cov - Compute covariance matrix per class
  • mean - Compute class mean vectors
  • log - Natural logarithm of data matrix
  • log2 - Base 2 logarithm of data matrix
  • log10 - Base 10 logarithm of data matrix

3. sdlab class: Handling labels, decisions and indexed properties ↩

  • sdlab - Create sdlab object
  • length - Get number of entries
  • getlist - Get label list object describing categories
  • getnames - Get string name for each label entry
  • getindices - Get index to label list for each label entry
  • getsizes - Get number of entries per category (i.e. samples per class)
  • getpriors - Get relative fraction os entries in each category (i.e class priors)
  • sdrelab - Rename classes or define meta-classes

4. sdlist class: List of categories ↩

  • sdlist - Create list object
  • getnames - Get category names
  • length - Get number of categories
  • ind2name - Convert category index to name
  • name2ind - Convert name to category index
  • isname - Test if a category is present

5. sdops class: Sets of operating points; making decisions ↩

  • sdops - Create an operting point or a set of operating points
  • decide - Perform decisions on data set with classifier soft outputs
  • getlist - Get list of decision categories
  • getdata - Get data of operating points (thresholds or weights)
  • subset - Get a subset of operating points by indices

6. sdppl class: Pipelines and classifiers ↩

6.1. Constructing and using pipelines ↩

  • sdppl - Low level pipeline constructor
  • sdconvert - Convert externally trained algorithms into pipelines (PRTools, DD_Tools)
  • getlab - Get pipeline labels (feature labels of the output data set)
  • getlist - Get pipeline decision list
  • getoutput - Return pipeline output type (decision, similarity, distance, ...)
  • sdexe - Execute pipeline on data used the libPRSD MEX library
  • sdexport - Export pipeline for execution out of Matlab using libPRSD.dll

6.2. Dimensionality reduction and data representation ↩

  • sdpca - Principal Component Analysis (PCA)
  • sdlda - Linear Discriminant Analysis (Fisher projection)
  • sdprox - Construction of proxmity representation (distances or similarities to prototypes)
  • sdscale - Data scaling
  • sdnorm - Normalization of soft-outputs
  • sdfeatsel - Feature selection

6.3. Classifiers (models) ↩

  • sdnmean - Nearest mean classifier
  • sdlinear - Linear discriminant assuming normal densities
  • sdquadratic - Quadratic discriminant assuming normal densities
  • sdgauss - Gaussian model
  • sdmixture - Gaussian mixture automatically estimated number of components
  • sdfisher - Fisher linear discriminant (LDA + sdlinear)
  • sdknn - k-th nearest neighbor classifier
  • sdkmeans - k-means classifier
  • sdkcentres - k-centres classifier
  • sdparzen - Parzen classifier
  • sdneural - Feed-forward neural network
  • sdnbayes - Naive Bayes classifier
  • sdtree - Decision tree classifier
  • sdsvc - Support vector classifier

6.4. High-level classifier tools ↩

  • sddetector - Detector for any model based on ROC analysis or one-class approach
  • sdcascade - Classifier cascade or hierarchy of classifiers
  • sdreject - Adding reject option to a trained pipeline
  • sdstackgen - Stacked generalization for building trained combiners (produces unbiased model output on the training set by cross-validation)

7. sdroc class: ROC analysis and evaluation ↩

7.1. Performing ROC analysis; accessing operating points and performance estimates ↩

  • sdroc - Performing ROC analysis on classifier outputs
  • getops - Return operating points in the sdroc object
  • getdata - Get data (performance estimates) stored in the sdroc object
  • sddrawroc - Interactive ROC plot

7.2. Selecting a specific operating point ↩

  • getcurop - Get index of the current operating point
  • setcurop - Set the current operating point by index or performance
  • subset - Select a subset of operating points
  • constrain - Select a subset of operating point by applying performance constraints
  • sddecide - Add a default operating point to a trained classifier

7.3. Evaluation and performance testing ↩

  • sdtest - Estimating error/performance of a classifier
  • sdconfmat - Estimate confusion matrices from decisions or classifer output
  • sdconfmatind - Get indices of samples falling into a specific confusion matrix cell
  • sdloss - Compute loss based on confusion matrices
  • sdcrossval - Cross-validation over samples or a user-defined property

8. sdalg class: Custom algorithms ↩

  • sdalg - Construct algorithm object
  • istrained - Test if the algorithm is trained
  • sda_pca_clf - Algorithm example: PCA dim.reduction followed by a classifier
  • sda_pca_clf_roc - Algorithm example: PCA dim.reduction + classifier + set operating point by ROC analysis
(c) 2007-2011, PR Sys Design