Data table functions

R/utils.R

@@ -494,3 +494,81 @@ screen_forbidden <- function(fn) {
   }
   rslt
 }
+
+#' fast_group_by
+#'
+#' A version of group_by that uses data.table instead of dplyr. Creates groups, runs a user specified function, ungroups and returns
+#' a processed tibble. Please use this function for grouping only numeric data.
+#'
+#' This group_by function that uses data.table offers a much higher speed
+#' especially when working with high volume datasets. This function can also be
+#' called within dplyr pipes. data.table will also inherently ensure consistency
+#' between LHS and RHS. The function will perform a combination of a group_by , mutate and ungroup.
+#'
+#' Example-
+#'
+#' A group_by with dplyr - grouped_data <- data -> group_by(iso,year,glu_code) -> mutate(value=sum(value)) -> ungroup()
+#'
+#' Same group_by with data.table - grouped_data <- fast_group_by(data, by=c("iso","year","glu_code"),colname = "value", func = "sum" )
+#'
+#' @param df The tibble on which the group_by is to be performed
+#' @param by A vector of strings with the criteria for the group_by.
+#' @param colname A string with the column name on which the grouping is to be performed
+#' @param func A string with the function to be performed. Default is set to "sum"
+#' @return A tibble with the aggregated data.
+#' @importFrom data.table as.data.table
+#' @importFrom tibble as_tibble
+#' @importFrom dplyr %>%
+#' @author kbn 24 Mar 2020
+#' @export
+fast_group_by<- function(df,by,colname="value",func= "sum"){
+
+
+    #Convert relevant column to numeric
+    df[,colname]<- as.numeric(df[,colname])
+
+    #Store as data.table
+    df <- as.data.table(df)
+
+    #Complete operations
+    df<- df[, (colname) := (get(func)(get(colname))), by]
+
+    #Save back to tibble
+    df<- as_tibble(df)
+
+    return(df)
+}
+
+#' data_table_bind
+#'
+#' A binding function that uses data.table. This can be used as a replacement for rbind or bind_rows.
+#'
+#' This binding function takes advantage of the data processing capabilities of data.table. This can be
+#' called within dplyr pipes.
+#'
+#'Example-
+#'
+#'Bind 2 datasets (x,y) with same columns using the following,
+#'
+#'Bound_dataset<- data_table_bind(x,y)
+#'
+#' @param ... The tibbles to be merged.
+#' @importFrom data.table as.data.table rbindlist
+#' @importFrom tibble as_tibble
+#' @return A tibble with combined data.
+#' @author kbn 24 Mar 2020
+#' @export
+data_table_bind<-function(...){
+
+    #Create a list for binding
+    list_for_bind =list(...)
+
+    #bind into one dataframe using rbindlist
+    df <- rbindlist(list_for_bind,use.names=TRUE)
+
+    #Return as tibble
+    df<-as_tibble(df)
+
+    return(df)
+
+}