R CMD check OK
[talweg.git] / 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 }