spine

SPINE Apply SPINE to raw data in one call. [scores, Result] = spine(data, measure, orientation) [scores, Result] = spine(data, measure, orientation, 'Name', Value, ...) [scores, Result, Info] = spine(...) MEASURE and ORIENTATION are both required. They are the scientific input: SPINE does not know what your data means, and choosing them is the modelling step. Neither has a default. Accepted data numeric vector one value per object numeric matrix rows are objects, columns are features struct array one element per object; numeric scalar fields become the features, a text field becomes the identifiers table numeric variables become the features, row names or a text variable become the identifiers Measure Any name accepted by spineMeasure. Type "help spineMeasure" for the menu. Orientation "upper" large measure is atypical "lower" small measure is atypical "bilateral" atypical on either side of an anchor object Options 'Objects' "rows" (default) or "columns", for numeric matrices 'K' neighbour count for the knn measure (default 5) 'ObjectIds' identifiers, overriding anything found in the data Any remaining option is passed to spineOptions, so BoundaryMode, ProfileWeights, and BilateralAnchor all work here unchanged. Outputs scores one score per object, in YOUR object order Result the readable record, everything in your object order (below) Info the reference record returned by spineCore, left untouched Result fields scores one score per object, 0 typical to 1 most atypical data the ingested data, rows as objects, for plotting mu the computed measure, one value per object stratum stratum index per object nStrata number of strata nBoundaries number of cuts, always nStrata - 1 boundaryValues measure value at each cut, largest value below the cut boundaryIntervals the open interval each cut lies in, one row per cut strataRanges measure range of each stratum, one row per stratum objectIds the identifiers used measure what spineMeasure did: name, parameters, dropped columns orientation the orientation you chose order, inverse the sort mapping applied before calling spine. When objects were omitted, order holds the caller's row numbers of the kept objects in ascending measure order, and inverse is NaN at the omitted rows. Info the reference record, for spinePlot and for auditing boundaryValues is the transferable result. Info.boundaries are ranks into an internally sorted copy and mean nothing outside this call; the measure VALUE at a cut is what you apply to new data or quote in a paper. Because the measure is sorted before spineCore sees it, the index-valued fields of Info (permutation, sortedIds, boundaryLeftObjectIds, boundaryRightObjectIds) refer to that sorted copy, not to your ordering. Result holds the remapped view. This file sits at the top of the MATLAB folder and needs no setup. Calling it puts core and interface on the path for you. It calls SPINE only through spineCore(mu, orientation, options) and spineOptions. No file under core is read or modified. Author Abd AlRahman AlMomani Assistant Professor of Data Science and Mathematics, ERAU, AZ almomaniar@gmail.com