fact
- Data management
- cdel — removes columns from a Div structure
- centering — centers a Div structure
- csv2div — loads a csv file to the Div format
- csv2div2 — loads to the Div format a csv file with: field separator=comma ',' and decimal separator = dot '.'
- dcsv2div — loads a csv file containing only numerical values without missing values and organized as a matrix
- dcsv2div2 — loads a csv file containing only numerical values without missing values and organized as a matrix with field separator = comma ',' and decimal separator = dot '.'
- div — converts an entry into a Div structure
- div2csv — exports a Div structure to the csv format
- div2dcsv — exports to the csv format, without labels, a matrix or the .d field within a Div structure
- div2mat — exports a Div structure or a structure of Divs to the .mat format
- div2tab — exports a Div structure to tabular-csv
- groupcreate — yields groups organized as a list of Div structures from a conjunctive or disjunctive codification
- indexseek — in a vector of numbers, seeks the index coresponding to the value which is the closer to a given number
- isconj — verifies if a column vector x (or x.d from a Div structure)
corresponds to a conjunctive codification
- isdisj — verifies if a matrice x (or x.d from a Div structure) corresponds to a disjunctive classification
- islist — verifies if an input is a list and converts to Div the elements which were not Div structures previously
- jstr — justifies a vector of strings to a number of characters equal to (ndigits+1)
- nandel — removing of lines or columns containing missing values (NaN)
- num2str — transforms a vector of numbers into a vector of strings which are justified for a number of digits equal to ndigits
- rdel — removes lines
- reorder — sorts two Div structures from the identifiers they share
- scsv2sci — loads a .csv file containing strings organized as a matrix; column separator=semi colon(;)
- standardize — divides each column by its standard deviation
- str2conj — yields a conjunctive vector of classes identified within strings
- strseek — seeks a string within the elements of a vector
- Formats conversions
- div2lmata — exports a Div structure into a lmata text file
- lmata2div — imports a lmata text file into a Div structure
- mz_hdf5tosci — import in Scilab of a HDF5-format file obtained by MSconvert from a raw-format mass spectrum
- Statistics
- arand_index — calculates the adjusted Rand index for two classifications
- corrmat — matrix of the correlation coefficients between the column vectors of two matrices
- covmat — matrix of covariances between the column vectors of two matrices
- snk — analysis of variance et comparison of means by Student-Newman-Keuls (SNK)
- std — calculates the standard deviation of each column of x
- wilks — calculates the Wilks lambda = inter-classes inertia / total inertia
- (X) One matrix of observations/variables
- cspcana — principal component analysis on centered-standardized data
- ica_blocs_signals — the number of independent components is determined from the correlations between blocs issued from the dataset
- ica_dwresiduals — the number of independent components in ICA is determined from the Durbin-Watson criterium
- icajade — yields a matrix for extracting the independent components using the Jade algorithm
- icascores — extracts independent components (signals, loadings) and proportions (scores) using the icajade function
- kcmeans — classifies the observations into groups according to the k-means classification
- outlier — calculation of 3 parameters: T2-Hotelling, Q or residual variances, and leverage, useful to identify outliers
- pcaapply — applies a PCA model, calculated with a first dataset, to a second dataset
- pcana — principal component analysis
- (X) One matrix of contingency
- ca — correspondence analysis
- (X->X) Pretraitments by orthogonal projection
- detrending — applies a Detrend correction by orthogonal projection to a Vandermonde matrix
- dop — calculates the eigenvectors for orthogonal projection with Dynamic Orthogonal Projection
- epo — calculates the eigenvectors and the parameters for tuning External Parameter Orthogonalization (EPO) with the classical approach based on the means for each influence factor
- epoe — calculates the eigenvectors and the parameters for tuning External Parameter Orthogonalization (EPO) with a centering by sample inspired from EROS
- eros — calculates the eigenvectors and the parameters for tuning Error Removal by Orthogonal Substraction (EROS)
- iirp — calculates the eigenvectors and the parameters for tuning Independent Interference Reduction (IIR)
- osc — computes the parameters of orthogonal signal correction (OSC) according to the direct method
- pop_dextract — extracts the matrix of detrimental information with one of these methods: IIR ('iirp'), EPO ('epo'), EROS('eros' ou 'epoe')
- pop_dtune — calculates the tuning parameters of the TOP-Transfer by Orthogonal Projection method, for which the variability is issued from two measurements on the same samples (ex: with two devices)
- pop_dtune — calculates the tuning parameters of an orthogonal projection for which the matrix of detrimental information is given
- popapply — applies a model obtained with 'epo','epoe','eros','iirp' or 'pop_dtune' to a matrix xtest or a test dataset (xtest,ytest)
- vanderm — calculates a Vandermonde matrix for the Detrend correction
- (X->X) Pretraitments by variable selection
- covsel — selection of variables by maximizing their covariance to y
- Peaks alignment
- obiwarp — correction of retention times in liquid and gaz chromatography
- (X->X) Other pretaitments
- gapremove — corects from sharp baseline shifts, typically observed when lamps are switched
- msc — performs a MSC-multiplicative scatter correction on a set of spectra
- normc — sets to 1 the norms of the columns; columns of 0 are unchanged
- pds_apply — application of a PDS model to a new dataset obtained with a spectrometer B
- pds_calc — calculation of PDS-piecewise direct standardization, a method which uses the spectra acquired with a spectrometer B (master) to estimate the spectra that would have been obtained with a spectrometer A (slave)
- savgol — smoothing, first or second derivate of spectra according to Savitzsky-Golay
- simufilters — conversion of a wavelength unit into another wavelength unit, often shifted, according to the equation: x_new = x_old * filter'; very useful to convert frequencies into wavelengths, and vice-versa
- snv — standard normal variate (SNV)
- (X->y) Linear calibrations
- covsel_mlr — variable selection using covsel, then multiple linear regression (MLR)
- dimension_tune — 8 criteria to tune the dimension of a PLSR or PCR, obtained on the predictions (RMSECV), on the b-coefficients (b-norm, Durbin-Watson, morphological factor and Gini), and on the deflated X-matrices (KMO-Kaiser-Meyer-Olkin, VIF-Variance Inflation Factor and Durbin-Watson)
- ikpls — partial least squares regression computed with the improved-kernel algorithm
- mlr — multiple linear regression (MLR)
- pcr — principal components regression (PCR)
- pls — partial least squares regression, projection to latent structure regression or PLSR computed with the standard algorithm
- regapply — applies a calibration model calculated with pls, vodka, pcr, mlr or ridge to a new dataset
- ridge — the Ridge regression
- spls — stacked PLS regression
- vodka — the Vodka regression
- (X->y) Non-linear calibrations
- nns_buildbayes — building of a neural network with 1 hidden layer and with bias
- nns_init — initialization of a neural network with 1 hidden layer, with bias
- nns_simul — predictions by a neural network with 1 hidden layer and with bias
- nns_simulbis — prediction and prediction errors of a neural network with 1 hidden layer and with bias
- nns_simulter — prediction errors by a neural network with 1 hidden layer and with bias
- i-Asca-anova
- asca — ASCA - analysis of variance simultaneous components analysis
- (X->Y_classes) Classification
- copda — classification by orthogonal projection (COP) with cross-validation
- covsel_fda — variable selection using Covsel, then factorial discriminant analysis on the selected variables
- daaply — application of a discriminant model to a test dataset
- fda — factorial discriminant analysis (FDA) with cross-validation
- forwda — forward discriminant analysis with cross-validation
- knnda — k-near-neighbors with cross-validation
- plsda — PLS discriminant analysis (PLS-DA) with cross-validation: a PLS2 (Simpls) computes scores from the disjonctive matrix of the classes, then each observation is attributed to a class using these scores
- plsfda — PLS factoriel discriminant analysis (PLS-FDA) with cross-validation: a PLS2 (Simpls) computes scores from the disjonctive matrix of the classes, then a factoriel discriminant analysis is applied to the scores, yielding new scores; each observation is attributed to a class using these new scores
- (X1-X2…Xk) Multi-way analysis
- acom1 — common components and co-inertia analysis
- ccswa — common components and specific weights analysis
- comdim — common components and specific weights analysis, algorithmm different from ccswa
- statis — the Statis method
- statislda — Statis-linear discriminant analysis method
- Graphics
- barycentermap — cloud of points; all the observations of the same class are represented by the same colour and are connected to the barycenter of their class
- coloredmap — colored figure representing observations by the identifier of their class
- corrplot — correlations circle between the variables of one or more matrices on the one hand; and two axes representing orthogonal scores (e.g. from a PCA or a PLSR) on the other hand
- curves — plot of the column-vectors of a Div structure
- dendro — calculates distances between observations, then plots the classification tree as a dendrogram
- diacmap — figure representing the observations by a triangle whose color depends on the class of the observation
- dotcmap — figure representing the observations by a dot whose color depends on the class of the observation
- holdoff — a new figure is created for each new graph of Fact
- holdon — the last figure is used for representing each new graph of Fact
- kcmap — black and white figure representing observations by the identifier of their class
- kscmap — black and white figure representing the observations by a symbol (or a number if more than 7 classes) different for each class
- map — black and white figure representing observations by their labels
- regplot — plots reference values vs predicted values and additional informations: R2, bias, RMSE, slope and offset
- scmap — figure representing the observations by a symbol (or a number if more than 7 classes) whose color is different for each class
- starcmap — figure representing the observations by a star whose color depends on the class of the observation