pkg/R/getData.R

   1 #' getData
   2 #'
   3 #' Acquire data as a Data object; see ?Data.
   4 #'
   5 #' Since series are given in columns (database format), this function builds series one
   6 #' by one and incrementally grows a Data object which is finally returned.
   7 #'
   8 #' @param ts_data Time-series, as a data frame (DB style: 2 columns, first is date/time,
   9 #'   second is value) or a CSV file.
  10 #' @param exo_data Exogenous variables, as a data frame or a CSV file; first column is
  11 #'   dates, next block are measurements for the day, and final block are exogenous
  12 #'   forecasts (for the same day).
  13 #' @param input_tz Timezone in the input files ("GMT" or e.g. "Europe/Paris")
  14 #' @param date_format How date/time are stored (e.g. year/month/day hour:minutes;
  15 #'   see ?strptime)
  16 #' @param working_tz Timezone to work with ("GMT" or e.g. "Europe/Paris")
  17 #' @param predict_at When does the prediction take place? Integer, in hours. Default: 0
  18 #' @param limit Number of days to extract (default: Inf, for "all")
  19 #'
  20 #' @return An object of class Data
  21 #'
  22 #' @examples
  23 #' ts_data = read.csv(system.file("extdata","pm10_mesures_H_loc.csv",package="talweg"))
  24 #' exo_data = read.csv(system.file("extdata","meteo_extra_noNAs.csv",package="talweg"))
  25 #' data = getData(ts_data, exo_data, limit=120)
  26 #' @export
  27 getData = function(ts_data, exo_data, input_tz="GMT", date_format="%d/%m/%Y %H:%M",
  28     working_tz="GMT", predict_at=0, limit=Inf)
  29 {
  30     # Sanity checks (not full, but sufficient at this stage)
  31     if (!is.character(input_tz) || !is.character(working_tz))
  32         stop("Bad timezone (see ?timezone)")
  33     input_tz = input_tz[1]
  34     working_tz = working_tz[1]
  35     if ( (!is.data.frame(ts_data) && !is.character(ts_data)) ||
  36             (!is.data.frame(exo_data) && !is.character(exo_data)) )
  37         stop("Bad time-series / exogenous input (data frame or CSV file)")
  38     if (is.character(ts_data))
  39         ts_data = ts_data[1]
  40     if (is.character(exo_data))
  41         exo_data = exo_data[1]
  42     predict_at = as.integer(predict_at)[1]
  43     if (predict_at<0 || predict_at>23)
  44         stop("Bad predict_at (0-23)")
  45     if (!is.character(date_format))
  46         stop("Bad date_format (character)")
  47     date_format = date_format[1]
  48
  49     ts_df =
  50         if (is.character(ts_data))
  51             read.csv(ts_data)
  52         else
  53             ts_data
  54     # Convert to the desired timezone (usually "GMT" or "Europe/Paris")
  55     formatted_dates_POSIXlt = strptime(as.character(ts_df[,1]), date_format, tz=input_tz)
  56     ts_df[,1] = format(
  57         as.POSIXct(formatted_dates_POSIXlt, tz=input_tz), tz=working_tz, usetz=TRUE)
  58
  59     exo_df =
  60         if (is.character(exo_data))
  61             read.csv(exo_data)
  62         else
  63             exo_data
  64     # Times in exogenous variables file are ignored: no conversions required
  65
  66     line = 1 #index in PM10 file (24 lines for 1 cell)
  67     nb_lines = nrow(ts_df)
  68     nb_exos = ( ncol(exo_df) - 1 ) / 2
  69     data = Data$new()
  70     i = 1 #index of a cell in data
  71     while (line <= nb_lines)
  72     {
  73         time = c()
  74         serie = c()
  75         repeat
  76         {
  77             {
  78                 time = c(time, ts_df[line,1])
  79                 serie = c(serie, ts_df[line,2])
  80                 line = line + 1
  81             };
  82             if (line >= nb_lines + 1
  83                 || as.POSIXlt(ts_df[line-1,1],tz=working_tz)$hour == predict_at)
  84             {
  85                 break
  86             }
  87         }
  88
  89         exo = as.data.frame( exo_df[i,2:(1+nb_exos)] )
  90         exo_hat =
  91             if (i < nrow(exo_df))
  92                 as.data.frame( exo_df[i+1,(1+nb_exos+1):(1+2*nb_exos)] )
  93             else
  94                 NA #exogenous prediction for next day are useless on last day
  95         data$append(time, serie, exo, exo_hat)
  96         if (i >= limit)
  97             break
  98         i = i + 1
  99     }
 100     if (length(data$getCenteredSerie(1)) < length(data$getCenteredSerie(2)))
 101         data$removeFirst()
 102     if (length(data$getCenteredSerie(data$getSize()))
 103         < length(data$getCenteredSerie(data$getSize()-1)))
 104     {
 105         data$removeLast()
 106     }
 107     data
 108 }