v1.6.3

DESCRIPTION

@@ -1,11 +1,10 @@
 Package: DSWE
 Title: Data Science for Wind Energy
-Version: 1.6.1
+Version: 1.6.3
 Authors@R: c(
     person(given = "Nitesh", family = "Kumar", role = c("aut"), email = "nitesh.kumar@tamu.edu"),
-    person(given = "Abhinav", family = "Prakash", role = c("aut"), email = "abhinavp@tamu.edu"),
+    person(given = "Abhinav", family = "Prakash", role = c("aut"), email = "abhinavp@aggienetwork.com"),
     person(given = "Yu", family = "Ding", role = c("aut","cre"), email = "yuding@tamu.edu"),
-    person(given = "Rui", family = "Tuo", role = c("ctb","cph"), email = "ruituo@tamu.edu"),
     person(given = "Effi", family = "Latiffianti", role = c("ctb","cph"), email = "latiffianti@tamu.edu"))
 Description: Data science methods used in wind energy applications. 
               Current functionalities include creating a multi-dimensional power curve model, 
@@ -26,7 +25,7 @@ URL: https://github.com/TAMU-AML/DSWE-Package, https://aml.engr.tamu.edu/book-ds
 BugReports: https://github.com/TAMU-AML/DSWE-Package/issues
 Encoding: UTF-8
 LazyData: true
-RoxygenNote: 7.1.2
+RoxygenNote: 7.2.3
 LinkingTo: 
     Rcpp (>= 1.0.4.6) ,
     RcppArmadillo (>= 0.9.870.2.0)

NAMESPACE

@@ -37,4 +37,4 @@ importFrom(stats,predict)
 importFrom(stats,sd)
 importFrom(stats,var)
 importFrom(utils,write.table)
-useDynLib(DSWE)
+useDynLib(DSWE, .registration = TRUE)

R/AMKGapSubroutines.R

@@ -22,11 +22,21 @@
 
 get.dpill = function (cov, y)
 {
-  bw <- KernSmooth::dpill(cov, y)
+  bw <- try(
+    expr = KernSmooth::dpill(cov, y), silent = TRUE
+  )
+  if (inherits(bw, "try-error")){
+    bw = sd(cov)
+  }
   if (is.nan(bw)) {
     par <- 0.06
     while (is.nan(bw)) {
-      bw <- KernSmooth::dpill(cov, y, proptrun = par)
+      bw <- try(
+        expr = KernSmooth::dpill(cov, y, proptrun = par), silent = TRUE
+      )
+      if (inherits(bw, "try-error")){
+        bw = sd(cov)
+      }
       par <- par + 0.01
     }
   }
@@ -172,7 +182,7 @@ bw.gap = function (y, x, id.dir = NA, id.adp = id.dir) {
               0 & cutpt[[p]][length(cutpt[[p]])] >= 360)
             cov.sub[which(cov.sub < cutpt[[p]][2])] <- cov.sub[which(cov.sub <
                                                                        cutpt[[p]][2])] + 360
-          KernSmooth::dpill(cov.sub, y[id.bin])
+          get.dpill(cov.sub, y[id.bin])
         })
       })
       list(bw.fix = bw.fix, bw.adp = bw.adp, id.adp = id.adp,

R/AMKSubroutines.R

@@ -129,7 +129,14 @@ calculateWeights = function(trainX,testpoint,bandwidth,nMultiCov,fixedCov,cirCov
 computeBandwidth = function(trainY,trainX,cirCov){
   bandwidth = rep(0,ncol(trainX))
   for (i in 1:ncol(trainX)){
-    bandwidth[i] = KernSmooth::dpill(trainX[,i],trainY)
+    bw = try(
+      expr = KernSmooth::dpill(trainX[,i],trainY), silent=TRUE
+    )
+    if (inherits(bw, "try-error")) {
+      bandwidth[i] = sd(trainX[,i])
+    } else {
+      bandwidth[i] = bw
+    }
   }
   if(all(!is.na(cirCov))){
   for (i in cirCov){

R/CovMatchSubroutines.R

@@ -21,7 +21,7 @@
 # SOFTWARE.
 
 #' @importFrom  matrixStats colMins colMaxs colSds
-#' @useDynLib DSWE
+#' @useDynLib DSWE, .registration = TRUE
 #' @importFrom Rcpp sourceCpp
 
 CovMatch.Mult = function(dname, cov, wgt, cov.circ){

R/KnnPCFit.R

@@ -46,7 +46,9 @@
 #' knn_model = KnnPCFit(data, xCol, yCol, subsetSelection)
 #' 
 #' @export
+#' @useDynLib DSWE, .registration = TRUE
 #' @importFrom FNN knn.reg knnx.index
+#' 
 KnnPCFit = function(data, xCol, yCol, subsetSelection = FALSE){
 
   if(!is.matrix(data) & !is.data.frame(data)){

R/RcppExports.R

@@ -2,34 +2,34 @@
 # Generator token: 10BE3573-1514-4C36-9D1C-5A225CD40393
 
 computeWeightedY <- function(X, y, params) {
-    .Call('_DSWE_computeWeightedY', PACKAGE = 'DSWE', X, y, params)
+    .Call(`_DSWE_computeWeightedY`, X, y, params)
 }
 
 predictGP <- function(X, weightedY, Xnew, params) {
-    .Call('_DSWE_predictGP', PACKAGE = 'DSWE', X, weightedY, Xnew, params)
+    .Call(`_DSWE_predictGP`, X, weightedY, Xnew, params)
 }
 
 computeLogLikGP_ <- function(X, y, params) {
-    .Call('_DSWE_computeLogLikGP_', PACKAGE = 'DSWE', X, y, params)
+    .Call(`_DSWE_computeLogLikGP_`, X, y, params)
 }
 
 computeLogLikGradGP_ <- function(X, y, params) {
-    .Call('_DSWE_computeLogLikGradGP_', PACKAGE = 'DSWE', X, y, params)
+    .Call(`_DSWE_computeLogLikGradGP_`, X, y, params)
 }
 
 computeLogLikGradGPZeroMean_ <- function(X, y, params) {
-    .Call('_DSWE_computeLogLikGradGPZeroMean_', PACKAGE = 'DSWE', X, y, params)
+    .Call(`_DSWE_computeLogLikGradGPZeroMean_`, X, y, params)
 }
 
 computeDiffCov_ <- function(X1, y1, X2, y2, XT, theta, sigma_f, sigma_n, beta) {
-    .Call('_DSWE_computeDiffCov_', PACKAGE = 'DSWE', X1, y1, X2, y2, XT, theta, sigma_f, sigma_n, beta)
+    .Call(`_DSWE_computeDiffCov_`, X1, y1, X2, y2, XT, theta, sigma_f, sigma_n, beta)
 }
 
 computeConfBand_ <- function(diffCovMat, confLevel) {
-    .Call('_DSWE_computeConfBand_', PACKAGE = 'DSWE', diffCovMat, confLevel)
+    .Call(`_DSWE_computeConfBand_`, diffCovMat, confLevel)
 }
 
 matchcov <- function(ref, obj, thres, circ_pos, flag) {
-    .Call('_DSWE_matchcov', PACKAGE = 'DSWE', ref, obj, thres, circ_pos, flag)
+    .Call(`_DSWE_matchcov`, ref, obj, thres, circ_pos, flag)
 }
 

R/funGPSubroutines.R

@@ -20,7 +20,7 @@
 # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 # SOFTWARE.
 
-#' @useDynLib DSWE
+#' @useDynLib DSWE, .registration = TRUE
 #' @importFrom Rcpp sourceCpp
 #' 
 estimateParameters= function(datalist, covCols, yCol, opt_method, limitMemory, optimSize, rngSeed){

R/tempGP.R

@@ -1,6 +1,6 @@
 # MIT License
 # 
-# Copyright (c) 2020 Abhinav Prakash, Rui Tuo, and Yu Ding
+# Copyright (c) 2020-2022 Abhinav Prakash and Yu Ding
 # 
 # Permission is hereby granted, free of charge, to any person obtaining a copy
 # of this software and associated documentation files (the "Software"), to deal
@@ -25,6 +25,7 @@
 #' @param trainX A matrix with each column corresponding to one input variable. 
 #' @param trainY A vector with each element corresponding to the output at the corresponding row of \code{trainX}.
 #' @param trainT A vector for time indices of the data points. By default, the function assigns natural numbers starting from 1 as the time indices. 
+#' @param max_thinning_number An integer specifying the max lag to compute the thinning number. If the PACF does not become insignificant till \code{max_thinning_number}, then \code{max_thinning_number} is used for thinning.
 #' @param fast_computation A Boolean that specifies whether to do exact inference or fast approximation. Default is \code{TRUE}.
 #' @param limit_memory An integer or \code{NULL}. The integer is used sample training points during prediction to limit the total memory requirement. Setting the value to \code{NULL} would result in no sampling, that is, full training data is used for prediction. Default value is \code{5000}.
 #' @param optim_control A list parameters passed to the Adam optimizer when \code{fast_computation} is set to \code{TRUE}. The default values have been tested rigorously and tend to strike a balance between accuracy and speed. \itemize{
@@ -82,6 +83,7 @@
 tempGP = function(trainX, trainY, trainT = NULL, 
                   fast_computation = TRUE,
                   limit_memory = 5000L,
+                  max_thinning_number = 20,
                   optim_control = list(batch_size = 100L, 
                                        learn_rate = 0.05, 
                                        max_iter = 5000L, 
@@ -131,6 +133,10 @@ tempGP = function(trainX, trainY, trainT = NULL,
       stop('trainX, trainY, and trainT must have the same number of data points.')
     }
 
+    if (!inherits(max_thinning_number, "integer")){
+      stop('max_thinning_number must be an integer.') 
+    }
+
     trainX = trainX[order(trainT),,drop = FALSE]
     trainY = trainY[order(trainT)]
     trainT = trainT[order(trainT)]
@@ -145,7 +151,7 @@ tempGP = function(trainX, trainY, trainT = NULL,
     limit_memory = as.integer(limit_memory)
   }
 
-  thinningNumber = computeThinningNumber(trainX)
+  thinningNumber = computeThinningNumber(trainX, max_thinning_number)
 
   if (thinningNumber > 0){
     thinnedBins = createThinnedBins(trainX,trainY,thinningNumber)  

R/tempGPSubroutines.R

@@ -1,6 +1,6 @@
 # MIT License
 # 
-# Copyright (c) 2020 Abhinav Prakash, Rui Tuo, and Yu Ding
+# Copyright (c) 2020-2022 Abhinav Prakash and Yu Ding
 # 
 # Permission is hereby granted, free of charge, to any person obtaining a copy
 # of this software and associated documentation files (the "Software"), to deal
@@ -20,9 +20,17 @@
 # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 # SOFTWARE.
 
-computeThinningNumber = function(trainX){
-  thinningNumber = max(apply(trainX,2,function(col) 
-    min(which(c(1,abs(stats::pacf(col, plot = FALSE)$acf[,1,1])) <= (2/sqrt(nrow(trainX)))))))
+#' @useDynLib DSWE, .registration = TRUE
+
+computeThinningNumber = function(trainX, max_thinning_number){
+  thinning_vec = rep(max_thinning_number, ncol(trainX))
+  for (col_idx in c(1:length(thinning_vec))){
+    col_thinning_vec = which(c(1,abs(stats::pacf(trainX[,col_idx], plot = FALSE, lag.max = max_thinning_number)$acf[,1,1])) <= (2/sqrt(nrow(trainX))))
+    if (length(col_thinning_vec) != 0){
+      thinning_vec[col_idx] = min(col_thinning_vec)
+    }
+  }
+  thinningNumber = max(thinning_vec)
   return(thinningNumber)
 }