tsImpute: Time Series Imputation

In jeffwong/imputation: imputation

Description

Time Series Imputation using Boosted Trees Fill each column by treating it as a regression problem. For each column i, use boosted regression trees to predict i using all other columns except i. If the predictor variables also contain missing data, the gbm function will itself use surrogate variables as substitutes for the predictors. This imputation function can handle both categorical and numeric data.

Usage

  tsImpute(time, dimension, metric, max.iters = 2,
    cv.fold = 2, n.trees = 100, verbose = T, ...)

Arguments

time	a vector of dates or datetime objects
dimension	a data frame of exogenous predictor variables
metric	a matrix where each column represents a time series
max.iters	number of times to iterate through the columns and impute each column with fitted values from a regression tree
cv.fold	number of folds that gbm should use internally for cross validation
n.trees	the number of trees used in gradient boosting machines
verbose	if TRUE print status updates
...	additional params passed to gbm

Examples

dates = timeSequence(from = '2012-01-01', to = '2012-12-31', by = 'day')
  dimensions = sample(c("A", "B"), 366, replace = TRUE)
  numA = length(which(dimensions == "A")); numB = length(which(dimensions == "B"))
  metrics = matrix(0, 366, 2)
  metrics[which(dimensions == "A"),1] = rnorm(numA, mean=1)
  metrics[which(dimensions == "A"),2] = rnorm(numA, mean=5)
  metrics[which(dimensions == "B"),1] = rnorm(numB, mean=-10)
  metrics[which(dimensions == "B"),2] = rnorm(numB, mean=-5)
  tp = projectDate(as.Date(dates))
  monday.indices = which(tp$weekday == "Monday")
  metrics[sample(monday.indices, 20),] = NA
  tsImpute(as.Date(dates), dimensions, metrics)

jeffwong/imputation documentation built on May 19, 2019, 4:02 a.m.