• The STANDARD functions toolbox
  
  
  CLUSTER
  
  
  Hopkins:- To evaluate the clustering tendency using Hopkins statistic (and Forina modification)
  
  
  INSPECT
  
  
  Randtest:- Comparing the predictive accuracy of models using a simple randomisation test
  
  Rms:- Calculates Root Mean Squared Error (RMSE), PRedicted REsidual Sum of Squared errors (PRESS), Standard Error (SE) and bias for a vector of predicted y-values
  
  
  MISSING
  
  
  Inimiss:- Substitution of missing elements by the means of variables
  
  Mmean:- Calculation of mean(X) for X with missing elements
  
  Mpls:- PLS model for Y,X, where X contains missing elements
  
  Mstd:- Calculation of std of X with missing elements
  
  Msvd:- svd(X) with missing data
  
  Msvdcv:- svd(X) with missing data and cross-validation
  
  Mvar:- Calculation of variance of X, for X with missing elements
  
  
  OUTLIER
  
  
  Gru_dix:- To detect atypical objects using the Grubbs or Dixon outlier test applied to Principal Component scores (t).
  
  Hadi:- Outlier detection based on clean subset selection method proposed by Hadi.
  
  Hotelor:- Hotellings T2 chart. Classical test for small number of original variables.
  
  Mahcal:- To calculate the Mahalanobis distance in the original space for CALIBRATION samples
  
  Rao:- To calculate the Rao statistics from the data (spectral)matrix X.
  This file should be used before running rao_gru.m since the output Dsq is the input of rao_gru.m.
  
  Rao_gru:- To detect atypical objects using the Rao statistic and the modified Grubbs test.
  
  
  PCA
  
  
  Kpca:- Fast PCA with efficient EVD algorithm
  
  Sfa:- Significant factor analysis - A program designed to help determine the number of significant factors in a matrix
  
  Svd2:- Faster singular value decomposition
  
  
  POTFUNC
  
  
  Centpf:- Compute the potential of the centroid of different pairs of points in the calibration set to determine the optimal smoothing value
  
  Criticv:- This program is to decide the cut-off (potential) value (alpha confidential level), calculate the alpha-error (of the training set), and detect the outliers which are wrongly rejected.
  
  Gauspf:- Estimation of the probability density with a Gaussian potential function
  
  Loopf:- Leave-one-out cross-validation approach, to determine the optimal value of the smoothing parameter
  
  Potenf:- Use potential functions method, which based on the distance methods, to predict the test samples (to detect prediction outliers).
  
  Smoothpf:- To calculate the smoothing value of potential function.
  
  Trianpf:- Estimation of the probability density with a Triangular potential function
  
  
  PREPROC
  
  
  Center:- Centering along columns, rows or double centering
  
  Detrend:- De-trending (Baseline correction: To remove the effect of baseline-shift and curvilinearity (in the case of densely packed solid samples, e.g. powders)
  
  Invcente:- Inverse centering
  
  Invrange:- Returns a range-scaled data set to its original scale
  
  Invscale:- Restores a scaled data set to its original mean and variance
  
  Msc:- Multiple Scatter Correction:To remove the effect of physical light scatter from the spectrum. (Compensation for particle size effects.)
  
  Offset:- To correct (spectra) for a baseline shift
  
  Range:- Each element in the data set is scaled to a range determined by the user
  
  Scale:- Performs scaling to unit variance or autoscaling (centering + scaling to unit variance)
  
  Snv:- Standard Normal Variate Transformation Row centering, followed by row scaling
  
  Splbase:- Spline baseline correction for NIR spectra
  
  
  SELECT
  
  
  Duplex:- Duplex algorithm for sample subset selection in a data set
  
  Kenston:- Kennard-Stone design (e.g. for subset selection)
  
  Randsel:- Random subset selection
  
  Ranksel:- Subset selection according to ranked y-values
  
  
  SIGPROC
  
  
  Deriv:- Derivative computation by using the Savitsky-Golay algorithm
  
  Expsmoot:- Performs exponential smoothing on spectral data
  
  Ff:- Filtering in Fourier domain by using 1) Hard thresholding on the Fourier coefficients 2) Windowing the Fourier transform of the signal
  
  Genfact:- Derivative computation by using the Savitsky-Golay algorithm: Weight computation
  
  Grampoly:- Derivative computation by using the Savitsky-Golay algorithm: Weight computation
  
  Hfour:- Filtering in Fourier domain by using Hard thresholding on the Fourier coefficients
  
  Lowp:- Filtering in Fourier domain by using windowing the Fourier transform of the signal
  
  Smooth:- Performs average smoothing on spectral data
  
  Wave:- Coefficients of Quadrature Mirror Filter (QMF)
  
  Weight:- Derivative computation by using the Savitsky-Golay algorithm: Weight computation
  
  
  
  
  
  

  Creation	Sébastien Gourvénec	11-Oct-02
  Last Update	Sébastien Gourvénec	11-Oct-02
  	hits on this page since	25-Jan-02
  	hits on all pages since	22-May-97