[talweg.git] / pkg / R / F_Neighbors2.R

#' @include Forecaster.R
#'
#' Neighbors2 Forecaster
#'
#' Predict tomorrow as a weighted combination of "futures of the past" days.
#' Inherits \code{\link{Forecaster}}
#'
Neighbors2Forecaster = R6::R6Class("Neighbors2Forecaster",
	inherit = Forecaster,

	public = list(
		predictShape = function(data, today, memory, horizon, ...)
		{
			# (re)initialize computed parameters
			private$.params <- list("weights"=NA, "indices"=NA, "window"=NA)

			# Do not forecast on days with NAs (TODO: softer condition...)
			if (any(is.na(data$getCenteredSerie(today))))
				return (NA)

			# Determine indices of no-NAs days followed by no-NAs tomorrows
			fdays = getNoNA2(data, max(today-memory,1), today-1)

			# Get optional args
			simtype = ifelse(hasArg("simtype"), list(...)$simtype, "mix") #or "endo", or "exo"
			kernel = ifelse(hasArg("kernel"), list(...)$kernel, "Gauss") #or "Epan"
			if (hasArg(h_window))
			{
				return ( private$.predictShapeAux(data,
					fdays, today, horizon, list(...)$h_window, kernel, simtype, TRUE) )
			}

			# Indices of similar days for cross-validation; TODO: 20 = magic number
			cv_days = getSimilarDaysIndices(today, data, limit=20, same_season=FALSE, days_in=fdays)

			# Function to optimize h : h |--> sum of prediction errors on last 45 "similar" days
			errorOnLastNdays = function(h, kernel, simtype)
			{
				error = 0
				nb_jours = 0
				for (i in seq_along(cv_days))
				{
					# mix_strategy is never used here (simtype != "mix"), therefore left blank
					prediction = private$.predictShapeAux(data,
						fdays, cv_days[i], horizon, h, kernel, simtype, FALSE)
					if (!is.na(prediction[1]))
					{
						nb_jours = nb_jours + 1
						error = error +
							mean((data$getCenteredSerie(cv_days[i]+1)[1:horizon] - prediction)^2)
					}
				}
				return (error / nb_jours)
			}

			if (simtype != "endo")
			{
				h_best_exo = optimize(
					errorOnLastNdays, c(0,7), kernel=kernel, simtype="exo")$minimum
			}
			if (simtype != "exo")
			{
				h_best_endo = optimize(
					errorOnLastNdays, c(0,7), kernel=kernel, simtype="endo")$minimum
			}

			if (simtype == "endo")
			{
				return (private$.predictShapeAux(data,
					fdays, today, horizon, h_best_endo, kernel, "endo", TRUE))
			}
			if (simtype == "exo")
			{
				return (private$.predictShapeAux(data,
					fdays, today, horizon, h_best_exo, kernel, "exo", TRUE))
			}
			if (simtype == "mix")
			{
				h_best_mix = c(h_best_endo,h_best_exo)
				return(private$.predictShapeAux(data,
					fdays, today, horizon, h_best_mix, kernel, "mix", TRUE))
			}
		}
	),
	private = list(
		# Precondition: "today" is full (no NAs)
		.predictShapeAux = function(data, fdays, today, horizon, h, kernel, simtype, final_call)
		{
			fdays_cut = fdays[ fdays < today ]
			# TODO: 3 = magic number
			if (length(fdays_cut) < 3)
				return (NA)

			# Neighbors: days in "same season"; TODO: 60 == magic number...
			fdays = getSimilarDaysIndices(today, data, limit=60, same_season=TRUE, days_in=fdays_cut)
			if (length(fdays) <= 1)
				return (NA)
			levelToday = data$getLevel(today)
			distances = sapply(fdays, function(i) abs(data$getLevel(i)-levelToday))
			#TODO: 2, 3, 5, 10 magic numbers here...
			dist_thresh = 2
			min_neighbs = min(3,length(fdays))
			repeat
			{
				same_pollution = (distances <= dist_thresh)
				nb_neighbs = sum(same_pollution)
				if (nb_neighbs >= min_neighbs) #will eventually happen
					break
				dist_thresh = dist_thresh + 3
			}
			fdays = fdays[same_pollution]
			max_neighbs = 10
			if (nb_neighbs > max_neighbs)
			{
				# Keep only max_neighbs closest neighbors
				fdays = fdays[ sort(distances[same_pollution],index.return=TRUE)$ix[1:max_neighbs] ]
			}
			if (length(fdays) == 1) #the other extreme...
			{
				if (final_call)
				{
					private$.params$weights <- 1
					private$.params$indices <- fdays
					private$.params$window <- 1
				}
				return ( data$getSerie(fdays[1])[1:horizon] ) #what else?!
			}

			if (simtype != "exo")
			{
				h_endo = ifelse(simtype=="mix", h[1], h)

				# Distances from last observed day to days in the past
				serieToday = data$getSerie(today)
				distances2 = sapply(fdays, function(i) {
					delta = serieToday - data$getSerie(i)
					mean(delta^2)
				})

				sd_dist = sd(distances2)
				if (sd_dist < .Machine$double.eps)
				{
#					warning("All computed distances are very close: stdev too small")
					sd_dist = 1 #mostly for tests... FIXME:
				}
				simils_endo =
					if (kernel=="Gauss")
						exp(-distances2/(sd_dist*h_endo^2))
					else
					{
						# Epanechnikov
						u = 1 - distances2/(sd_dist*h_endo^2)
						u[abs(u)>1] = 0.
						u
					}
			}

			if (simtype != "endo")
			{
				h_exo = ifelse(simtype=="mix", h[2], h)

				M = matrix( nrow=1+length(fdays), ncol=1+length(data$getExo(today)) )
				M[1,] = c( data$getLevel(today), as.double(data$getExo(today)) )
				for (i in seq_along(fdays))
					M[i+1,] = c( data$getLevel(fdays[i]), as.double(data$getExo(fdays[i])) )

				sigma = cov(M) #NOTE: robust covariance is way too slow
				# TODO: 10 == magic number; more robust way == det, or always ginv()
				sigma_inv =
					if (length(fdays) > 10)
						solve(sigma)
					else
						MASS::ginv(sigma)

				# Distances from last observed day to days in the past
				distances2 = sapply(seq_along(fdays), function(i) {
					delta = M[1,] - M[i+1,]
					delta %*% sigma_inv %*% delta
				})

				sd_dist = sd(distances2)
				if (sd_dist < .25 * sqrt(.Machine$double.eps))
				{
#					warning("All computed distances are very close: stdev too small")
					sd_dist = 1 #mostly for tests... FIXME:
				}
				simils_exo =
					if (kernel=="Gauss")
						exp(-distances2/(sd_dist*h_exo^2))
					else
					{
						# Epanechnikov
						u = 1 - distances2/(sd_dist*h_exo^2)
						u[abs(u)>1] = 0.
						u
					}
			}

			similarities =
				if (simtype == "exo")
					simils_exo
				else if (simtype == "endo")
					simils_endo
				else #mix
					simils_endo * simils_exo
			similarities = similarities / sum(similarities)

			prediction = rep(0, horizon)
			for (i in seq_along(fdays))
				prediction = prediction + similarities[i] * data$getCenteredSerie(fdays[i]+1)[1:horizon]

			if (final_call)
			{
				private$.params$weights <- similarities
				private$.params$indices <- fdays
				private$.params$window <-
					if (simtype=="endo")
						h_endo
					else if (simtype=="exo")
						h_exo
					else #mix
						c(h_endo,h_exo)
			}

			return (prediction)
		}
	)
)
Commit	Line	Data
5c49f6ce BA	1	#' @include Forecaster.R
	2	#'
	3	#' Neighbors2 Forecaster
	4	#'
	5	#' Predict tomorrow as a weighted combination of "futures of the past" days.
	6	#' Inherits \code{\link{Forecaster}}
	7	#'
	8	Neighbors2Forecaster = R6::R6Class("Neighbors2Forecaster",
	9	inherit = Forecaster,
	10
	11	public = list(
	12	predictShape = function(data, today, memory, horizon, ...)
	13	{
	14	# (re)initialize computed parameters
	15	private$.params <- list("weights"=NA, "indices"=NA, "window"=NA)
	16
	17	# Do not forecast on days with NAs (TODO: softer condition...)
	18	if (any(is.na(data$getCenteredSerie(today))))
	19	return (NA)
	20
	21	# Determine indices of no-NAs days followed by no-NAs tomorrows
9db234c5	22	fdays = getNoNA2(data, max(today-memory,1), today-1)
5c49f6ce BA	23
5c49f6ce BA	24	# Get optional args
5e838b3e	25	simtype = ifelse(hasArg("simtype"), list(...)$simtype, "mix") #or "endo", or "exo"
5c49f6ce BA	26	kernel = ifelse(hasArg("kernel"), list(...)$kernel, "Gauss") #or "Epan"
	27	if (hasArg(h_window))
	28	{
	29	return ( private$.predictShapeAux(data,
5e838b3e	30	fdays, today, horizon, list(...)$h_window, kernel, simtype, TRUE) )
5c49f6ce BA	31	}
5c49f6ce BA	32
6774e53d BA	33	# Indices of similar days for cross-validation; TODO: 20 = magic number
6774e53d BA	34	cv_days = getSimilarDaysIndices(today, data, limit=20, same_season=FALSE, days_in=fdays)
5e838b3e	35
5c49f6ce	36	# Function to optimize h : h \|--> sum of prediction errors on last 45 "similar" days
5e838b3e	37	errorOnLastNdays = function(h, kernel, simtype)
5c49f6ce BA	38	{
	39	error = 0
	40	nb_jours = 0
5e838b3e	41	for (i in seq_along(cv_days))
5c49f6ce BA	42	{
5c49f6ce BA	43	# mix_strategy is never used here (simtype != "mix"), therefore left blank
5e838b3e BA	44	prediction = private$.predictShapeAux(data,
5e838b3e BA	45	fdays, cv_days[i], horizon, h, kernel, simtype, FALSE)
5c49f6ce BA	46	if (!is.na(prediction[1]))
	47	{
	48	nb_jours = nb_jours + 1
	49	error = error +
6774e53d	50	mean((data$getCenteredSerie(cv_days[i]+1)[1:horizon] - prediction)^2)
5c49f6ce BA	51	}
	52	}
	53	return (error / nb_jours)
	54	}
	55
5e838b3e BA	56	if (simtype != "endo")
	57	{
	58	h_best_exo = optimize(
ea5c7e56	59	errorOnLastNdays, c(0,7), kernel=kernel, simtype="exo")$minimum
5e838b3e BA	60	}
	61	if (simtype != "exo")
	62	{
	63	h_best_endo = optimize(
ea5c7e56	64	errorOnLastNdays, c(0,7), kernel=kernel, simtype="endo")$minimum
5e838b3e BA	65	}
	66
	67	if (simtype == "endo")
	68	{
	69	return (private$.predictShapeAux(data,
	70	fdays, today, horizon, h_best_endo, kernel, "endo", TRUE))
	71	}
	72	if (simtype == "exo")
	73	{
	74	return (private$.predictShapeAux(data,
	75	fdays, today, horizon, h_best_exo, kernel, "exo", TRUE))
	76	}
	77	if (simtype == "mix")
	78	{
	79	h_best_mix = c(h_best_endo,h_best_exo)
	80	return(private$.predictShapeAux(data,
	81	fdays, today, horizon, h_best_mix, kernel, "mix", TRUE))
	82	}
5c49f6ce BA	83	}
	84	),
	85	private = list(
	86	# Precondition: "today" is full (no NAs)
5e838b3e	87	.predictShapeAux = function(data, fdays, today, horizon, h, kernel, simtype, final_call)
5c49f6ce	88	{
6774e53d	89	fdays_cut = fdays[ fdays < today ]
5c49f6ce	90	# TODO: 3 = magic number
6774e53d	91	if (length(fdays_cut) < 3)
5c49f6ce BA	92	return (NA)
5c49f6ce BA	93
6774e53d BA	94	# Neighbors: days in "same season"; TODO: 60 == magic number...
	95	fdays = getSimilarDaysIndices(today, data, limit=60, same_season=TRUE, days_in=fdays_cut)
	96	if (length(fdays) <= 1)
ee8b1b4e	97	return (NA)
9db234c5	98	levelToday = data$getLevel(today)
6774e53d	99	distances = sapply(fdays, function(i) abs(data$getLevel(i)-levelToday))
ea5c7e56 BA	100	#TODO: 2, 3, 5, 10 magic numbers here...
	101	dist_thresh = 2
	102	min_neighbs = min(3,length(fdays))
6774e53d	103	repeat
5c49f6ce	104	{
6774e53d	105	same_pollution = (distances <= dist_thresh)
ea5c7e56 BA	106	nb_neighbs = sum(same_pollution)
ea5c7e56 BA	107	if (nb_neighbs >= min_neighbs) #will eventually happen
6774e53d	108	break
ea5c7e56	109	dist_thresh = dist_thresh + 3
5c49f6ce	110	}
6774e53d	111	fdays = fdays[same_pollution]
ea5c7e56 BA	112	max_neighbs = 10
	113	if (nb_neighbs > max_neighbs)
	114	{
	115	# Keep only max_neighbs closest neighbors
	116	fdays = fdays[ sort(distances[same_pollution],index.return=TRUE)$ix[1:max_neighbs] ]
	117	}
	118	if (length(fdays) == 1) #the other extreme...
a866acb3 BA	119	{
	120	if (final_call)
	121	{
	122	private$.params$weights <- 1
6774e53d	123	private$.params$indices <- fdays
a866acb3 BA	124	private$.params$window <- 1
a866acb3 BA	125	}
6774e53d	126	return ( data$getSerie(fdays[1])[1:horizon] ) #what else?!
a866acb3 BA	127	}
a866acb3 BA	128
5e838b3e BA	129	if (simtype != "exo")
	130	{
	131	h_endo = ifelse(simtype=="mix", h[1], h)
9db234c5	132
5e838b3e BA	133	# Distances from last observed day to days in the past
5e838b3e BA	134	serieToday = data$getSerie(today)
6774e53d	135	distances2 = sapply(fdays, function(i) {
5e838b3e BA	136	delta = serieToday - data$getSerie(i)
	137	mean(delta^2)
	138	})
5c49f6ce	139
5e838b3e BA	140	sd_dist = sd(distances2)
	141	if (sd_dist < .Machine$double.eps)
	142	{
5c49f6ce	143	# warning("All computed distances are very close: stdev too small")
5e838b3e BA	144	sd_dist = 1 #mostly for tests... FIXME:
	145	}
	146	simils_endo =
	147	if (kernel=="Gauss")
	148	exp(-distances2/(sd_dist*h_endo^2))
	149	else
	150	{
	151	# Epanechnikov
	152	u = 1 - distances2/(sd_dist*h_endo^2)
	153	u[abs(u)>1] = 0.
	154	u
	155	}
5c49f6ce	156	}
5e838b3e BA	157
	158	if (simtype != "endo")
	159	{
	160	h_exo = ifelse(simtype=="mix", h[2], h)
	161
6774e53d	162	M = matrix( nrow=1+length(fdays), ncol=1+length(data$getExo(today)) )
5e838b3e	163	M[1,] = c( data$getLevel(today), as.double(data$getExo(today)) )
6774e53d BA	164	for (i in seq_along(fdays))
6774e53d BA	165	M[i+1,] = c( data$getLevel(fdays[i]), as.double(data$getExo(fdays[i])) )
5e838b3e BA	166
5e838b3e BA	167	sigma = cov(M) #NOTE: robust covariance is way too slow
ee8b1b4e BA	168	# TODO: 10 == magic number; more robust way == det, or always ginv()
ee8b1b4e BA	169	sigma_inv =
6774e53d	170	if (length(fdays) > 10)
ee8b1b4e BA	171	solve(sigma)
	172	else
	173	MASS::ginv(sigma)
a866acb3	174
5e838b3e	175	# Distances from last observed day to days in the past
6774e53d	176	distances2 = sapply(seq_along(fdays), function(i) {
5e838b3e BA	177	delta = M[1,] - M[i+1,]
	178	delta %% sigma_inv %% delta
	179	})
	180
	181	sd_dist = sd(distances2)
	182	if (sd_dist < .25 * sqrt(.Machine$double.eps))
5c49f6ce	183	{
5e838b3e BA	184	# warning("All computed distances are very close: stdev too small")
5e838b3e BA	185	sd_dist = 1 #mostly for tests... FIXME:
5c49f6ce	186	}
5e838b3e BA	187	simils_exo =
	188	if (kernel=="Gauss")
	189	exp(-distances2/(sd_dist*h_exo^2))
	190	else
	191	{
	192	# Epanechnikov
	193	u = 1 - distances2/(sd_dist*h_exo^2)
	194	u[abs(u)>1] = 0.
	195	u
	196	}
	197	}
5c49f6ce	198
5e838b3e BA	199	similarities =
	200	if (simtype == "exo")
	201	simils_exo
	202	else if (simtype == "endo")
	203	simils_endo
	204	else #mix
	205	simils_endo * simils_exo
ea5c7e56	206	similarities = similarities / sum(similarities)
5c49f6ce BA	207
5c49f6ce BA	208	prediction = rep(0, horizon)
6774e53d BA	209	for (i in seq_along(fdays))
6774e53d BA	210	prediction = prediction + similarities[i] * data$getCenteredSerie(fdays[i]+1)[1:horizon]
5c49f6ce BA	211
	212	if (final_call)
	213	{
	214	private$.params$weights <- similarities
6774e53d	215	private$.params$indices <- fdays
5e838b3e BA	216	private$.params$window <-
	217	if (simtype=="endo")
	218	h_endo
	219	else if (simtype=="exo")
	220	h_exo
	221	else #mix
	222	c(h_endo,h_exo)
5c49f6ce BA	223	}
	224
	225	return (prediction)
	226	}
	227	)
	228	)