fix 2 tests out of 3; TODO: test forecasters
[talweg.git] / pkg / R / getData.R
... / ...
CommitLineData
1#' @title Acquire data in a clean format
2#'
3#' @description Take in input data frames and/or files containing raw data, and timezones, and
4#' output a Data object, roughly corresponding to a list where each cell contains all value
5#' for one day (see \code{?Data}). Current limitation: series (in working_tz) must start at
6#' right after midnight (to keep in sync with exogenous vars)
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 comlumn is dates,
11#' next block are measurements for the day, and final block are exogenous forecasts
12#' @param input_tz Timezone in the input files ("GMT" or e.g. "Europe/Paris")
13#' @param date_format How date/time are stored (e.g. year/month/day hour:minutes;
14#' see \code{strptime})
15#' @param working_tz Timezone to work with ("GMT" or e.g. "Europe/Paris")
16#' @param predict_at When does the prediction take place ? Integer, in hours. Default: 0
17#'
18#' @return An object of class Data
19#'
20#' @examples
21#' ts_data = read.csv(system.file("extdata","pm10_mesures_H_loc.csv",package="talweg"))
22#' exo_data = read.csv(system.file("extdata","meteo_extra_noNAs.csv",package="talweg"))
23#' getData(ts_data, exo_data, input_tz="Europe/Paris", working_tz="Europe/Paris", limit=150)
24#' @export
25getData = function(ts_data, exo_data, input_tz="GMT", date_format="%d/%m/%Y %H:%M",
26 working_tz="GMT", predict_at=0, limit=Inf)
27{
28 # Sanity checks (not full, but sufficient at this stage)
29 if (!is.character(input_tz) || !is.character(working_tz))
30 stop("Bad timezone (see ?timezone)")
31 input_tz = input_tz[1]
32 working_tz = working_tz[1]
33 if ( (!is.data.frame(ts_data) && !is.character(ts_data)) ||
34 (!is.data.frame(exo_data) && !is.character(exo_data)) )
35 stop("Bad time-series / exogenous input (data frame or CSV file)")
36 if (is.character(ts_data))
37 ts_data = ts_data[1]
38 if (is.character(exo_data))
39 exo_data = exo_data[1]
40 predict_at = as.integer(predict_at)[1]
41 if (predict_at<0 || predict_at>23)
42 stop("Bad predict_at (0-23)")
43 if (!is.character(date_format))
44 stop("Bad date_format (character)")
45 date_format = date_format[1]
46
47 ts_df =
48 if (is.character(ts_data))
49 read.csv(ts_data)
50 else
51 ts_data
52 exo_df =
53 if (is.character(exo_data))
54 read.csv(exo_data)
55 else
56 exo_data
57 # Convert to the desired timezone (usually "GMT" or "Europe/Paris")
58 formatted_dates_POSIXlt = strptime(as.character(ts_df[,1]), date_format, tz=input_tz)
59 ts_df[,1] = format(as.POSIXct(formatted_dates_POSIXlt), tz=working_tz, usetz=TRUE)
60
61 line = 1 #index in PM10 file (24 lines for 1 cell)
62 nb_lines = nrow(ts_df)
63 nb_exos = ( ncol(exo_df) - 1 ) / 2
64 data = list() #new("Data")
65 i = 1 #index of a cell in data
66 while (line <= nb_lines)
67 {
68 time = c()
69 serie = c()
70 repeat
71 {
72 {
73 time = c(time, ts_df[line,1])
74 serie = c(serie, ts_df[line,2])
75 line = line + 1
76 };
77 if (line >= nb_lines + 1 || as.POSIXlt(ts_df[line-1,1])$hour == predict_at)
78 break
79 }
80
81 # NOTE: if predict_at does not cut days at midnight, exogenous vars need to be shifted
82 exo_hat = as.data.frame( exo_df[
83 ifelse(predict_at>0,max(1,i-1),i) , (1+nb_exos+1):(1+2*nb_exos) ] )
84 exo = as.data.frame( exo_df[ ifelse(predict_at>0,max(1,i-1),i) , 2:(1+nb_exos) ] )
85 level = mean(serie, na.rm=TRUE)
86 centered_serie = serie - level
87 #data$append(time, centered_serie, level, exo_hat, exo_Jm1) #too slow; TODO: use R6 class
88 data[[length(data)+1]] = list("time"=time, "serie"=centered_serie, "level"=level,
89 "exo_hat"=exo_hat, "exo"=exo)
90 if (i >= limit)
91 break
92 i = i + 1
93 }
94 new("Data",data=data)
95}