projects / synclust.git / blame
| Commit | Line | Data |
|---|---|---|
| 15d1825d BA |
1 | #various util functions for the main program |
| 2 | ||
| 3 | #preliminary: replace NA's by averaging over each serie's values | |
| 4 | #TODO: find a better way to handle missing values | |
| 5 | replaceNAs = function(M) | |
| 6 | { | |
| 7 | n = nrow(M) | |
| 8 | m = ncol(M) | |
| 9 | res = M | |
| 10 | for (i in 1:n) | |
| 11 | { | |
| 12 | avg = mean(M[i,1:(m-2)] [!is.na(M[i,1:(m-2)])]) | |
| 13 | res[i,1:(m-2)] [is.na(M[i,1:(m-2)])] = avg | |
| 14 | } | |
| 15 | return (res) | |
| 16 | } | |
| 17 | ||
| 18 | #standardize matrix M (remove mean, divide by standard deviation) | |
| 19 | standardize = function(M) | |
| 20 | { | |
| 21 | avgM = colMeans(M, na.rm = TRUE) | |
| 22 | stdevs = sqrt( unlist( apply(M, 2, var, na.rm=TRUE) ) ) | |
| 23 | res = t(M) - avgM | |
| 24 | res = t(res / stdevs) | |
| 25 | return (list("M"=res,"avg"=avgM,"stv"=stdevs)) | |
| 26 | } | |
| 27 | ||
| 28 | #opposite of the previous function: get back M from standardized form | |
| 29 | destandardize = function(std) | |
| 30 | { | |
| 31 | M = std$M | |
| 32 | M = t(M) * std$stv | |
| 33 | M = t(M + std$avg) | |
| 34 | return (M) | |
| 35 | } | |
| 36 | ||
| 37 | #remap neighbors into some connex component | |
| 38 | remapNeighbors = function(NI, indices) | |
| 39 | { | |
| 40 | revIndices = rep(NA, length(NI)) | |
| 41 | nc = length(indices) | |
| 42 | for (ii in 1:nc) | |
| 43 | revIndices[ indices[ii] ] = ii | |
| 44 | locNI = list("ix"=as.list(rep(NA,nc)), "ds"=as.list(rep(NA,nc))) | |
| 45 | for (ii in 1:nc) | |
| 46 | { | |
| 47 | locNI$ix[[ii]] = revIndices[ NI$ix[[ indices[ii] ]] ] | |
| 48 | locNI$ds[[ii]] = NI$ds[[ indices[ii] ]] | |
| 49 | } | |
| 50 | return (locNI) | |
| 51 | } | |
| 52 | ||
| 53 | #check graph connexity | |
| 54 | getConnectedComponents = function(NIix) | |
| 55 | { | |
| 56 | return (.Call("getConnectedComponents", NIix)); | |
| 57 | } | |
| 58 | ||
| 59 | #auxiliary function to display clustering information | |
| 60 | promptForMapDisplay = function(stage, coordsM, NIix=NULL, clusters=NULL) | |
| 61 | { | |
| 62 | if (is.null(clusters)) | |
| 63 | clusters = rep(1, nrow(coordsM)) | |
| 64 | ||
| 65 | shouldDisplay = "" | |
| 66 | if (stage == "interm") | |
| 67 | shouldDisplay = readline(">>> show intermediate map of neighborhoods ? (y/n)\n") | |
| 68 | else if (stage == "final") | |
| 69 | { | |
| 70 | shouldDisplay = readline( | |
| 71 | ">>> show final map of clusters ? (y/n) \ | |
| 72 | NOTE: can be plotted later, see '? drawMapWithSites'\n") | |
| 73 | } | |
| 74 | ||
| 75 | if (shouldDisplay == "y") | |
| 76 | { | |
| 77 | drawMapWithSites(coordsM, clusters) | |
| 78 | if (!is.null(NIix)) | |
| 79 | drawNeighborhoodGraph(coordsM,NIix) | |
| 80 | print("Please press 'enter' to continue") | |
| 81 | readline() | |
| 82 | if (!is.null(dev.list())) | |
| 83 | dev.off() | |
| 84 | } | |
| 85 | } |