PRTools Contents

PRTools User Guide

feattypes

FEATTYPES

Determine feature types

[R,J,C,M,E] = feattypes(A) [R,J,C,M,E] = A*feattypes  [L1,L2, ...] = feattypes(A,T1,T2, ...] [L1,L2, ...] = A*feattypes(T1,T2, ...]   INPUT
    A Dataset
    T1,T2, ... Character strings specifying desired feature types,
                possible values 'R','J','J2','J3','C','C2','C3','M','E'.

Output
 R Indices of features with real values
 J Indices of integer features (J = [J2 J3])
 C Indices of categorical features (C = [C2 C3])
 L1,L2 ... Indices of features corresponding to T1, T2, ...

Description

This function sorts features in various types. This might be useful for  applying different classifiers for different feature types. Note that  missing values are coded in PRTools by a NaN. A simple procedure to  handle them is MISVAL.

- 'R' Real valued features
- 'J' Integer valued features
- 'J2' Binary (two-valued integer) features (not necessarily 0/1)
- 'J3' Integer features with more than two different valuers
- 'C' Categorical features
- 'C2' Categorical features with just two different values
- 'C3' Categorical features with more than two different values
- 'M' Features with only NaNs (all missing values)
- 'E' Features with the same value for all objects
- 'CM' Categorical features having the same values or NaN for all objects
- 'JM' Integer features having the same value or NaN for all objects
- 'RM' Real valued features having the same value or NaN for all objects

See also

datasets, mappings, featsel, setfeatdom, misval, cat2dset, cell2dset,

PRTools Contents

PRTools User Guide

This file has been automatically generated. If badly readable, use the help-command in Matlab.