1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
|
# --------------------------------------------------------------------
#
# ==> This function is just a CV wrapper to
# predict.st.matrix.ebird.ref.data!!!
# ==> Also new is the creation of an "index" file
# to hold all of the parameters needed to describe
# the st.matrix (file sequence)
#
#
# The ST Matrix:
# ------------------
# Basis for predictive experiments or ST analysis
# is predictions across a region in space and across.
# Often we want to know how spatial distrivution changed shrought time.
#
# Conceptually the ST Matrix as a data object is a set of
# prediction across a lattice in space and time.
# The st.matrix data object is a sequence of files
# Each one contains the predictions across a set of locations
# during a single time step.
# Each file in the sequence contains the predictions across
# the same set of locations evaluated at each of times
# in the temporal sequence.
#
# One set of ST.matrix files is written for each CV-fold
# model in cv.list
#
# The ST matrix can be thougth of as a source of psuedo-data
# with 3 added benefits:
# 1) Cleaned
# 2) Filled In
# 3) Uncertainty Estimates
#
#
# File Namining Conventions
# ---------------------------
#
#
# Naming Standards are being hardcoded for convenience/simplicity
# For example, one stem analysis per directory means that the
# STEM objects are uniquely identified by the directory name.
# So, I fixed the model names by hardcoding
# stem.model.tag <- "stem."
#
# st.matrix Names = paste(results.dir,st.matrix.file.tag,"st.matrix.",
# [sequence number], ".RData",sep="")
# These are kept in the CV result directories
# because they are CV specific.
#
# --------------------------------------------------------------------
# --------------------------------------------------------------------
# DMF 3.19.10 VT modifications
# * change formal parameter st.matrix.name, # st.matrix project name
predict.erd.st.matrix <- function(
stem.directory, # equivalent to stem.object
pred.data, #= srd.data,
cv.list, # subset of cv's
stem.predictor.names,
# --------------
st.matrix.name, # st.matrix project name
#st.matrix.directory,
jdate.seq,
year.seq,
conditioning.vars){ #= conditioning.vars)
# --------------------------------------------------------------------
stem.model.tag <- "stem." #hardcoded for convenience
print("**********")
print(stem.directory)
if ("/" == substring(stem.directory, nchar(stem.directory), nchar(stem.directory))) {
st.matrix.directory <- paste(stem.directory, st.matrix.name, "/", sep="")
stem.model.directory <- paste(stem.directory, "model", "/", sep = "")
}
else {
st.matrix.directory <- paste(stem.directory, "/", st.matrix.name, "/", sep="")
stem.model.directory <- paste(stem.directory, "/", "model", "/", sep="")
}
system(paste("mkdir ",st.matrix.directory, sep=""), intern=TRUE)
# ---------------------------
for (iii.cv in cv.list){
results.dir <- paste(st.matrix.directory, "st.matrix.",
as.character(iii.cv), sep = "")
system(paste("mkdir ",results.dir,sep=""), intern=TRUE)
# ------
model.dir <- paste(stem.model.directory, "stem.models.",
as.character(iii.cv), "/", sep="")
stem.ST.matrix.file.tag <-
paste(results.dir,"/", "st.matrix.",sep="")
## Load nfold specific parameter list
load(file = paste(model.dir,
"stem.model.par.list.", iii.cv, ".RData" , sep = "")) # D.stem
# ----------------------------
# Write ST.Matrix object to stem.results directories
# ----------------------------
print(names(pred.data$D.pred))
print(nrow(pred.data$D.pred))
predict.st.matrix.ebird.ref.data(
prediction.design = pred.data$D.pred,
prediction.design.locs = pred.data$locs,
# prediction.design.jdates, #not needed here
#--------------
model.predictor.names = stem.predictor.names,
model.filename = #stem.model$ensemble.model.filename,
paste(model.dir, stem.model.tag,sep=""),
ensemble.par.list = D.stem, #stem.model, #ensemble.par.list,
# --------------
save.name = stem.ST.matrix.file.tag,
# Temporal Sequence
jdate.seq =jdate.seq,
year.seq =year.seq,
# Conditioning Vars - Observation Process Design
conditioning.vars = conditioning.vars)
# ----------------------
# Save Index File
# ----------------------
st.matrix.list <- list(
stem.directory = stem.directory,
pred.data.locs = pred.data$locs,
cv.number=iii.cv, # subset of cv's
stem.predictor.names=stem.predictor.names,
# --------------
st.matrix.name = st.matrix.name,
jdate.seq = jdate.seq,
year.seq = year.seq,
conditioning.vars=conditioning.vars)
filename.tag <-
paste(results.dir,"st.matrix.list.",iii.cv,".RData",sep="")
save(st.matrix.list, file=filename.tag)
# ------------------
## KFW DF
## Local memory recovery block
rm(D.stem,st.matrix.list)
## KFW gc()
} # end iii.cv - data folds
# -----------------------------------------------------------------------------
# -----------------------------------------------------------------------------
# -----------------------------------------------------------------------------
## Clear the deck
rm(list = c("bs.rpart.maps", "conditioning.vars", "create.pd.grid",
"create.simple.STEM.ensemble", "D", "data.filename",
"D.erd.data.par.list", "D.erd.test.data.par.list",
"D.erd.train.data.par.list", "deviance.explained",
"ebirdReferenceDataFrame", "ensemble.partial.dependence",
"equal.wt.max.coverage.selection", "erdDataFile", "erdParameterListFile",
"exp.name", "fit.ensemble", "fit.stem", "halloween.maps", "iii.cv",
"iii.year", "initizalize.map.grid", "jdate.seq", "make.erd.data",
"make.functional.data.design.matrix", "make.SPAT.COVAR.pred.data",
"make.spatial.erd.prediction.design", "map.tag",
"maximum.coverage.selection", "model.dir", "monthly.maps", "n.intervals",
"n.intervals.per.year", "n.red", "plot.diagnostics",
"plot.STEM.temporal.design", "plot.ST.ensemble", "plot.vi.bdt",
"point.in.polygon", "point.in.polygon.contours", "point.in.shapefile",
"poisson.accuracy", "poisson.deviance", "poisson.pearson", "pop.rpart",
"pred.data", "predictive.performance", "predictors.index",
"predict.stem.ST.matrix", "predict.ST.ensemble", "predict.st.matrix",
"predict.st.matrix.ebird.ref.data", "protocol.tag", "resp.family",
"response.index", "response.truncation.cutoff", "results.dir", "return.list",
"rotate.ST.basis.pred", "rotate.ST.basis.sample",
"rpart.ensemble.diagnostics", "sample.ST.ensemble", "seasonal.window",
"smooth.st.predictions", "spatial.density.contours",
"spatial.performance.plot", "spp.common.name", "spp.dir", "spp.list",
"stateMapEnv", "stem.map.inference.split", "STEM.partial.dependence",
"stem.ST.matrix.file.tag", "st.ttt.pred", "summarize.ensemble",
"summary.1D.PD.ensemble", "surface.maps", "test.set.prediction.filename",
"ttsplit", "ttt.statement", "unique.locs.splitting",
"year.seq", "zzz.max"))
if (debugOutput == TRUE) {
rm(list = c("mem", "memoryMsg"))
}
## KFW gc()
# --------------------------------------------------------------------------
} #end function
# --------------------------------------------------------------------------
|
