AddMissingDates.R

AddMissingDates <- function(
            AGG_TS_DATA = FINAL_DATA[data_type == "TRAIN"],
            DATE_COLUMN = "date_ymd",
            TARGET_COLUMN = "visits",
            OTHER_COLUMNS = c("revenue_segment", "line", "serviceline_segment",
                              "finance_group", "finance_sub_group"),
            DEBUG = TRUE){

    if(DEBUG) message("AddMissingDates: Function Initialized  \n")

    if(!is.data.table(AGG_TS_DATA)){
      stop("AddMissingDates: The input data is not a data table  \n")
    }

    if(!class(AGG_TS_DATA[[DATE_COLUMN]]) %chin% c("Date")) {
      stop("AddMissingDates: The date column is not saved in date format  \n")
    }

    FUNCTION_OUTPUT <- list()

    ALL_FEATURES_TO_FILL <- unique(c(TARGET_COLUMN, OTHER_COLUMNS))

    if(all(ALL_FEATURES_TO_FILL %in% colnames(AGG_TS_DATA))){

        all_date_info <- AGG_TS_DATA[, .(min_date = min(AGG_TS_DATA[[DATE_COLUMN]]),
                                         max_date = max(AGG_TS_DATA[[DATE_COLUMN]]),
                                         unique_days = uniqueN(AGG_TS_DATA[[DATE_COLUMN]]),
                                         days_between = length(unique(seq.Date(min(AGG_TS_DATA[[DATE_COLUMN]]),
                                                                               max(AGG_TS_DATA[[DATE_COLUMN]]),
                                                                               by = "day"))))]

        all_date_info[, missing_dates := ifelse(unique_days == days_between, 0, 1)]

        if(all_date_info$missing_dates == 1){

            all_dates <- data.table(date_ymd = seq.Date(min(as.Date(all_date_info$min_date)),
                                                             max(as.Date(all_date_info$max_date)), by = "day"),
                                    value = NA)
            
            colnames(all_dates)[1] <- DATE_COLUMN
            # all_dates

            AGG_TS_DATA <- merge.data.table(all_dates, AGG_TS_DATA,
                                            by = DATE_COLUMN,
                                            all.x = TRUE)
            # AGG_TS_DATA

            AGG_TS_DATA[, visits := ifelse(is.na(get(TARGET_COLUMN)), 0.1, get(TARGET_COLUMN))]
            # AGG_TS_DATA

            FINAL_DATA <- AGG_TS_DATA[, mget(c(DATE_COLUMN, ALL_FEATURES_TO_FILL))]

            FUNCTION_OUTPUT[["FINAL_OUTPUT"]] <- FINAL_DATA

            if(!is.null(FUNCTION_OUTPUT) || !length(FUNCTION_OUTPUT) == 0){
              if(DEBUG) message("CreateFinalData: Data acquisition completed  \n ")
             
              return(FUNCTION_OUTPUT)
            } else {
              stop("CreateFinalData: Function returned an empty data set  \n")
            }
        }

    } else {

        stop("AddMissingDates: One or more input columns are missing from the dataset")

    }
}