fixes and improvements
[talweg.git] / R / getData.R
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 #' @export
21 getData = function(ts_data, exo_data,
22 input_tz="GMT", date_format="%d/%m/%Y %H:%M", working_tz="GMT", predict_at=0)
23 {
24 # Sanity checks (not full, but sufficient at this stage)
25 if (!is.character(input_tz) || !is.character(working_tz))
26 stop("Bad timezone (see ?timezone)")
27 input_tz = input_tz[1]
28 working_tz = working_tz[1]
29 if (!is.data.frame(ts_data) && !is.character(ts_data))
30 stop("Bad time-series input (data frame or CSV file)")
31 if (is.character(ts_data))
32 ts_data = ts_data[1]
33 predict_at = as.integer(predict_at)[1]
34 if (predict_at<0 || predict_at>23)
35 stop("Bad predict_at (0-23)")
36 if (!is.character(date_format))
37 stop("Bad date_format (character)")
38 date_format = date_format[1]
39
40 ts_df =
41 if (is.character(ts_data)) {
42 read.csv(ts_data)
43 } else {
44 ts_data
45 }
46 exo_df =
47 if (is.character(exo_data)) {
48 read.csv(exo_data)
49 } else {
50 exo_data
51 }
52 # Convert to the desired timezone (usually "GMT" or "Europe/Paris")
53 formatted_dates_POSIXlt = strptime(as.character(ts_df[,1]), date_format, tz=input_tz)
54 ts_df[,1] = format(as.POSIXct(formatted_dates_POSIXlt), tz=working_tz, usetz=TRUE)
55
56 line = 1 #index in PM10 file (24 lines for 1 cell)
57 nb_lines = nrow(ts_df)
58 nb_exos = ( ncol(exo_df) - 1 ) / 2
59 data = list() #new("Data")
60 i = 1 #index of a cell in data
61 while (line <= nb_lines)
62 {
63 time = c()
64 serie = c()
65 repeat
66 {
67 {
68 time = c(time, ts_df[line,1])
69 serie = c(serie, ts_df[line,2])
70 line = line + 1
71 };
72 if (line >= nb_lines + 1 || as.POSIXlt(ts_df[line-1,1])$hour == predict_at)
73 break
74 }
75
76 # NOTE: if predict_at does not cut days at midnight,
77 # for the exogenous to be synchronized they need to be shifted
78 if (predict_at > 0)
79 {
80 exo_hat = as.data.frame(exo_df[max(1,i-1),(1+nb_exos+1):(1+2*nb_exos)])
81 exo_Dm1 = if (i>=3) as.data.frame(exo_df[i-1,2:(1+nb_exos)]) else NA
82 } else
83 {
84 exo_hat = as.data.frame(exo_df[i,(1+nb_exos+1):(1+2*nb_exos)])
85 exo_Dm1 = if (i>=2) as.data.frame(exo_df[i-1,2:(1+nb_exos)]) else NA
86 }
87 i = i + 1
88 #center data
89 level = mean(serie, na.rm=TRUE)
90 centered_serie = serie - level
91 # data$append(time, centered_serie, level, exo_hat, exo_Jm1) #TODO: slow: why ?
92 data[[length(data)+1]] = list("time"=time, "serie"=centered_serie, "level"=level,
93 "exo_hat"=exo_hat, "exo_Dm1"=exo_Dm1)
94 }
95 new("Data",data=data)
96 }