Commit | Line | Data |
---|---|---|
e030a6e3 | 1 | #' @include Forecaster.R |
3d69ff21 | 2 | #' |
25b75559 | 3 | #' Neighbors Forecaster |
3d69ff21 | 4 | #' |
25b75559 BA |
5 | #' Predict tomorrow as a weighted combination of "futures of the past" days. |
6 | #' Inherits \code{\link{Forecaster}} | |
7 | NeighborsForecaster = R6::R6Class("NeighborsForecaster", | |
a66a84b5 | 8 | inherit = Forecaster, |
25b75559 BA |
9 | |
10 | public = list( | |
e030a6e3 | 11 | predictShape = function(today, memory, horizon, ...) |
3d69ff21 BA |
12 | { |
13 | # (re)initialize computed parameters | |
a66a84b5 | 14 | private$.params <- list("weights"=NA, "indices"=NA, "window"=NA) |
3d69ff21 | 15 | |
f17665c7 BA |
16 | # Get optional args |
17 | simtype = ifelse(hasArg("simtype"), list(...)$simtype, "mix") #or "endo", or "exo" | |
18 | kernel = ifelse(hasArg("kernel"), list(...)$kernel, "Gauss") #or "Epan" | |
19 | if (hasArg(h_window)) | |
a66a84b5 BA |
20 | { |
21 | return ( private$.predictShapeAux( | |
22 | fdays, today, horizon, list(...)$h_window, kernel, simtype, TRUE) ) | |
23 | } | |
3d69ff21 | 24 | |
3d69ff21 | 25 | # Determine indices of no-NAs days followed by no-NAs tomorrows |
f17665c7 BA |
26 | first_day = max(today - memory, 1) |
27 | fdays = (first_day:(today-1))[ sapply(first_day:(today-1), function(i) { | |
28 | !any(is.na(data$getSerie(i)) | is.na(data$getSerie(i+1))) | |
29 | }) ] | |
3d69ff21 | 30 | |
f17665c7 BA |
31 | # Indices of similar days for cross-validation; TODO: 45 = magic number |
32 | sdays = getSimilarDaysIndices(today, limit=45, same_season=FALSE) | |
3d69ff21 BA |
33 | |
34 | # Function to optimize h : h |--> sum of prediction errors on last 45 "similar" days | |
35 | errorOnLastNdays = function(h, kernel, simtype) | |
36 | { | |
37 | error = 0 | |
38 | nb_jours = 0 | |
f17665c7 | 39 | for (i in intersect(fdays,sdays)) |
3d69ff21 | 40 | { |
f17665c7 | 41 | # mix_strategy is never used here (simtype != "mix"), therefore left blank |
a66a84b5 | 42 | prediction = private$.predictShapeAux(fdays, i, horizon, h, kernel, simtype, FALSE) |
f17665c7 | 43 | if (!is.na(prediction[1])) |
3d69ff21 BA |
44 | { |
45 | nb_jours = nb_jours + 1 | |
f17665c7 | 46 | error = error + mean((data$getCenteredSerie(i+1)[1:horizon] - prediction)^2) |
3d69ff21 BA |
47 | } |
48 | } | |
49 | return (error / nb_jours) | |
50 | } | |
51 | ||
f17665c7 BA |
52 | if (simtype != "endo") |
53 | h_best_exo = optimize(errorOnLastNdays, c(0,10), kernel=kernel, simtype="exo")$minimum | |
3d69ff21 | 54 | if (simtype != "exo") |
f17665c7 | 55 | h_best_endo = optimize(errorOnLastNdays, c(0,10), kernel=kernel, simtype="endo")$minimum |
3d69ff21 BA |
56 | |
57 | if (simtype == "endo") | |
a66a84b5 | 58 | return(private$.predictShapeAux(fdays,today,horizon,h_best_endo,kernel,"endo",TRUE)) |
3d69ff21 | 59 | if (simtype == "exo") |
a66a84b5 | 60 | return(private$.predictShapeAux(fdays,today,horizon,h_best_exo,kernel,"exo",TRUE)) |
3d69ff21 BA |
61 | if (simtype == "mix") |
62 | { | |
f17665c7 | 63 | h_best_mix = c(h_best_endo,h_best_exo) |
a66a84b5 | 64 | return(private$.predictShapeAux(fdays,today,horizon,h_best_mix,kernel,"mix",TRUE)) |
3d69ff21 | 65 | } |
25b75559 BA |
66 | } |
67 | ), | |
68 | private = list( | |
3d69ff21 | 69 | # Precondition: "today" is full (no NAs) |
f17665c7 | 70 | .predictShapeAux = function(fdays, today, horizon, h, kernel, simtype, final_call) |
3d69ff21 | 71 | { |
f17665c7 | 72 | fdays = fdays[ fdays < today ] |
3d69ff21 | 73 | # TODO: 3 = magic number |
f17665c7 | 74 | if (length(fdays) < 3) |
3d69ff21 BA |
75 | return (NA) |
76 | ||
77 | if (simtype != "exo") | |
78 | { | |
79 | h_endo = ifelse(simtype=="mix", h[1], h) | |
80 | ||
81 | # Distances from last observed day to days in the past | |
f17665c7 BA |
82 | distances2 = rep(NA, length(fdays)) |
83 | for (i in seq_along(fdays)) | |
3d69ff21 | 84 | { |
25b75559 | 85 | delta = data$getCenteredSerie(today) - data$getCenteredSerie(fdays[i]) |
3d69ff21 BA |
86 | # Require at least half of non-NA common values to compute the distance |
87 | if (sum(is.na(delta)) <= 0) #length(delta)/2) | |
88 | distances2[i] = mean(delta^2) #, na.rm=TRUE) | |
89 | } | |
90 | ||
91 | sd_dist = sd(distances2) | |
99f83c9a BA |
92 | if (sd_dist < .Machine$double.eps) |
93 | sd_dist = 1 #mostly for tests... FIXME: | |
3d69ff21 | 94 | simils_endo = |
99f83c9a | 95 | if (kernel=="Gauss") |
3d69ff21 | 96 | exp(-distances2/(sd_dist*h_endo^2)) |
99f83c9a | 97 | else { #Epanechnikov |
3d69ff21 BA |
98 | u = 1 - distances2/(sd_dist*h_endo^2) |
99 | u[abs(u)>1] = 0. | |
100 | u | |
101 | } | |
102 | } | |
103 | ||
104 | if (simtype != "endo") | |
105 | { | |
106 | h_exo = ifelse(simtype=="mix", h[2], h) | |
107 | ||
25b75559 BA |
108 | M = matrix( nrow=1+length(fdays), ncol=1+length(data$getExo(today)) ) |
109 | M[1,] = c( data$getLevel(today), as.double(data$getExo(today)) ) | |
f17665c7 | 110 | for (i in seq_along(fdays)) |
25b75559 | 111 | M[i+1,] = c( data$getLevel(fdays[i]), as.double(data$getExo(fdays[i])) ) |
3d69ff21 BA |
112 | |
113 | sigma = cov(M) #NOTE: robust covariance is way too slow | |
613a986f | 114 | sigma_inv = solve(sigma) #TODO: use pseudo-inverse if needed? |
3d69ff21 BA |
115 | |
116 | # Distances from last observed day to days in the past | |
117 | distances2 = rep(NA, nrow(M)-1) | |
118 | for (i in 2:nrow(M)) | |
119 | { | |
120 | delta = M[1,] - M[i,] | |
121 | distances2[i-1] = delta %*% sigma_inv %*% delta | |
122 | } | |
123 | ||
124 | sd_dist = sd(distances2) | |
125 | simils_exo = | |
f17665c7 | 126 | if (kernel=="Gauss") |
3d69ff21 | 127 | exp(-distances2/(sd_dist*h_exo^2)) |
f17665c7 | 128 | else { #Epanechnikov |
3d69ff21 BA |
129 | u = 1 - distances2/(sd_dist*h_exo^2) |
130 | u[abs(u)>1] = 0. | |
131 | u | |
132 | } | |
133 | } | |
134 | ||
3d69ff21 | 135 | similarities = |
f17665c7 | 136 | if (simtype == "exo") |
3d69ff21 | 137 | simils_exo |
f17665c7 BA |
138 | else if (simtype == "endo") |
139 | simils_endo | |
140 | else #mix | |
141 | simils_endo * simils_exo | |
3d69ff21 BA |
142 | |
143 | prediction = rep(0, horizon) | |
a66a84b5 BA |
144 | for (i in seq_along(fdays)) |
145 | prediction = prediction + similarities[i] * data$getSerie(fdays[i]+1)[1:horizon] | |
3d69ff21 | 146 | prediction = prediction / sum(similarities, na.rm=TRUE) |
99f83c9a | 147 | |
3d69ff21 BA |
148 | if (final_call) |
149 | { | |
a66a84b5 BA |
150 | private$.params$weights <- similarities |
151 | private$.params$indices <- fdays | |
152 | private$.params$window <- | |
3d69ff21 BA |
153 | if (simtype=="endo") { |
154 | h_endo | |
155 | } else if (simtype=="exo") { | |
156 | h_exo | |
f17665c7 | 157 | } else { #mix |
3d69ff21 BA |
158 | c(h_endo,h_exo) |
159 | } | |
160 | } | |
99f83c9a | 161 | |
3d69ff21 BA |
162 | return (prediction) |
163 | } | |
164 | ) | |
165 | ) |