Fetching latest commit…
Cannot retrieve the latest commit at this time.
|Failed to load latest commit information.|
fvplay: Feature Vector Play This set of MATLAB scripts is intended for visualizing and preparing raw data for use with an SVM classifier. As is commonly known, the features that are used in the classifier are more important than the method itself, thus these tools provide an easy way to visualize and interpret data for best feature creation. Data is expected to be an n x m matrix of doubles with two classes. Package includes: fvprep fvplay fvplot %fvprep: Prepares data objects for use with rest of package. %%%%%%%%%% USAGE: fvprep(rawdata,labels) INPUT: rawdata: an n x m matrix of doubles, with n objects / subjects each with m features labels: a single column vector of size n x 1 with limited to values of -1 and 1 to distinguish the two classes OUTPUT: fv.features_vectorsraw data fv.index1 - indices in inputdata for label 1 (1) fv.index2 - indices in inputdata for label 2 (-1) fv.m - all means of features (columns) fv.u1, fv.s1, fv.v1: svd output for group 1 fv.u2, fv.s2, fv.v2: svd output for group 2 % fvplay takes as input an fv data object prepared by fvprep, and %%% allows the user to select a transform to plot (to explore the data) fvplot returns the data, and also displays plot in figures 1 and 2 USAGE: fvplot(fvobject,type) INPUT: fvobject: data object prepared by fvprep type: options include: quad: quadratic transformation exp: exponential transformation sig: sigmoid transformation log: logarithmic transformation OUTPUT: data: transformed n x m data matrix with n subjects (rows) and m features (columns) % fvplot takes as input the output of fvplay (an n x m matrix with n %% objects / subjects (rows) and m features (columns) and plots a mean subtracted histogram and heat map. USAGE: fvplot(fvobject,type) INPUT: transform: transformed n x m data matrix, with n objects and m features fv: fv data object OUTPUT: graphical... plots! % fvcorr takes as input a data matrix and associated labels (in format -1 and 1) and calculates the cross correlation matrix. USAGE: fvcorr(data, labels) INPUT: data: a n x M matrix with n subjects, m columns of features labels: a n X 1 column of labels, -1 and 1 OUTPUT: ccdata: cross correlation matrix