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
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
|
# test.parsnip.R: test the parsnip package with earth and other models
# Stephen Milborrow Sep 2020 Petaluma
source("test.prolog.R")
options(warn=1) # print warnings as they occur
library(earth)
cat("loading parsnip libraries\n") # these libraries take several seconds to load
library(tidymodels, quietly=TRUE, verbose=FALSE)
library(timetk)
library(lubridate)
cat("loaded parsnip libraries\n")
cat("parsnip version:", as.character(packageVersion("parsnip")[[1]]), "\n")
vdata <- data.frame(
resp = 1:23,
bool = c(F, F, F, F, F, T, T, T, T, T, T, T, T, F, F, T, T, T, T, T, T, T, T),
ord = ordered(c("ORD1", "ORD1", "ORD1",
"ORD1", "ORD1", "ORD1",
"ORD1", "ORD3", "ORD1",
"ORD2", "ORD2", "ORD2", "ORD2",
"ORD2", "ORD2", "ORD2",
"ORD3", "ORD3", "ORD3",
"ORD2", "ORD2", "ORD2", "ORD2"),
levels=c("ORD1", "ORD3", "ORD2")),
fac = as.factor(c("FAC1", "FAC1", "FAC1",
"FAC2", "FAC2", "FAC2",
"FAC3", "FAC1", "FAC1",
"FAC1", "FAC2", "FAC2", "FAC2",
"FAC2", "FAC2", "FAC2",
"FAC3", "FAC3", "FAC3",
"FAC1", "FAC3", "FAC3", "FAC3")),
str = c("STR1", "STR1", "STR1", # WILL BE TREATED LIKE A FACTOR
"STR1", "STR1", "STR1",
"STR2", "STR2", "STR2",
"STR3", "STR3", "STR2", "STR3",
"STR2", "STR3", "STR2",
"STR3", "STR3", "STR3",
"STR3", "STR3", "STR3", "STR3"),
num = c(1, 9, 2, 3, 14, 5, 6, 4, 5, 6.5, 3, 6, 5,
3, 4, 5, 6, 4, 5, 16.5, 3, 16, 15),
sqrt_num = sqrt(
c(1, 9, 2, 3, 14, 5, 6, 4, 5, 6.5, 3, 6, 5,
3, 4, 5, 6, 4, 5, 16.5, 3, 16, 15)),
int = c(1L, 1L, 3L, 3L, 4L, 4L, 3L, 5L, 3L, 6L, 7L, 8L, 10L,
13L, 14L, 3L, 13L, 5L, 13L, 16L, 17L, 18L, 11L),
date = as.Date(
c("2018-08-01", "2018-08-02", "2018-08-03",
"2018-08-04", "2018-08-05", "2018-08-06",
"2018-08-07", "2018-08-08", "2018-08-08",
"2018-08-10", "2018-08-10", "2018-08-11", "2018-08-11",
"2018-08-11", "2018-08-12", "2018-08-13",
"2018-08-10", "2018-08-15", "2018-08-17",
"2018-08-04", "2018-08-19", "2018-08-03", "2018-08-18")),
date_num = as.numeric(as.Date(
c("2018-08-01", "2018-08-02", "2018-08-03",
"2018-08-04", "2018-08-05", "2018-08-06",
"2018-08-07", "2018-08-08", "2018-08-08",
"2018-08-10", "2018-08-10", "2018-08-11", "2018-08-11",
"2018-08-11", "2018-08-12", "2018-08-13",
"2018-08-10", "2018-08-15", "2018-08-17",
"2018-08-04", "2018-08-19", "2018-08-03", "2018-08-18"))))
set.seed(2020)
splits <- initial_time_split(vdata, prop=.9)
#--- lm ----------------------------------------------------------------------
lm1 <- lm(resp~num+fac:int+date+ord+str, data=training(splits))
cat("lm1:\n")
print(summary(lm1))
set.seed(2020)
lmpar <- linear_reg(mode = "regression") %>%
set_engine("lm") %>%
fit(resp~num+fac:int+date+ord+str, data = training(splits))
stopifnot(identical(lm1$coeff, lmpar$fit$coeff))
predict.lm1 <- predict(lm1, testing(splits))
predict.lmpar <- lmpar %>% predict(testing(splits))
stopifnot(all(predict.lm1 == predict.lmpar))
par(mfrow = c(3, 3), mar = c(3, 3, 3, 1), mgp = c(1.5, 0.5, 0))
plotmo(lm1, do.par=2, SHOWCALL=TRUE)
plotres(lm1, which=c(3,1), do.par=FALSE)
plotmo(lmpar, do.par=2, SHOWCALL=TRUE)
plotres(lmpar, which=c(3,1), do.par=FALSE)
plotmo(lmpar$fit, do.par=2, SHOWCALL=TRUE)
plotres(lmpar$fit, which=c(3,1), do.par=FALSE)
par(org.par)
lmpar.sqrtnum <- linear_reg(mode = "regression") %>%
set_engine("lm") %>%
fit(resp~sqrt(num), data = training(splits))
#$$ TODO
# expect.err(try(plotmo(lmpar.sqrtnum)),
# "cannot get the original model predictors")
#--- earth -------------------------------------------------------------------
# note that sqrt(num) is ok, unlike parsnip models for lm and rpart
earth1 <- earth(resp~sqrt(num)+int+ord:bool+fac+str+date, degree=2,
data=training(splits), pmethod="none")
cat("earth1:\n")
print(summary(earth1))
set.seed(2020)
earthpar <- mars(mode = "regression", prune_method="none", prod_degree=2) %>%
set_engine("earth") %>%
fit(resp~sqrt(num)+int+ord:bool+fac+str+date, data = training(splits))
cat("earthpar:\n")
print(earthpar)
cat("summary(earthpar$fit)\n")
print(summary(earthpar$fit))
stopifnot(identical(earth1$coeff, earthpar$fit$coeff))
predict.earth1 <- predict(earth1, testing(splits))
predict.earthpar <- earthpar %>% predict(testing(splits))
stopifnot(all(predict.earth1 == predict.earthpar))
par(mfrow = c(3, 3), mar = c(3, 3, 3, 1), mgp = c(1.5, 0.5, 0))
plotmo(earth1, do.par=2, pt.col=3, SHOWCALL=TRUE)
set.seed(2020)
plotres(earth1, which=c(1,3), do.par=FALSE, pt.col=3, legend.pos="topleft")
par(org.par)
par(mfrow = c(3, 3), mar = c(3, 3, 3, 1), mgp = c(1.5, 0.5, 0))
plotmo(earthpar, do.par=2, pt.col=3, SHOWCALL=TRUE)
set.seed(2020)
plotres(earthpar, which=c(1,3), do.par=FALSE, pt.col=3, legend.pos="topleft")
par(org.par)
par(mfrow = c(3, 3), mar = c(3, 3, 3, 1), mgp = c(1.5, 0.5, 0))
plotmo(earthpar$fit, do.par=2, pt.col=3, SHOWCALL=TRUE)
set.seed(2020)
plotres(earthpar$fit, which=c(1,3), do.par=FALSE, pt.col=3, legend.pos="topleft")
par(org.par)
#--- rpart -------------------------------------------------------------------
library(rpart)
library(rpart.plot)
rpart1 <- rpart(resp~num+fac+int+date+ord+str, data=training(splits),
control=rpart.control(minsplit=1, cp=.0001))
cat("\nrpart.rules(rpart1)\n")
print(rpart.rules(rpart1))
set.seed(2020)
# TODO note need of model=TRUE below (needed only for further processing with e.g. plotmo)
rpartpar <- decision_tree(mode = "regression", min_n=1, cost_complexity=.0001) %>%
set_engine("rpart", model=TRUE) %>%
fit(resp~num+fac+int+date+ord+str, data = training(splits))
cat("\nrpart.rules(rpartpar$fit)\n")
print(rpart.rules(rpartpar$fit))
predict.rpart1 <- predict(rpart1, testing(splits))
predict.rpartpar <- rpartpar %>% predict(testing(splits))
stopifnot(all(predict.rpart1 == predict.rpartpar))
par(mfrow = c(3, 3), mar = c(3, 3, 3, 1), mgp = c(1.5, 0.5, 0))
plotmo(rpart1, do.par=2, SHOWCALL=TRUE, trace=0)
plotres(rpart1, which=c(3,1), do.par=FALSE)
plotmo(rpartpar, do.par=2, SHOWCALL=TRUE, trace=0)
plotres(rpartpar, which=c(3,1), do.par=FALSE)
plotmo(rpartpar$fit, do.par=2, SHOWCALL=TRUE)
plotres(rpartpar$fit, which=c(3,1), do.par=FALSE)
par(org.par)
# TODO note that this differs from the above rpart model in that we don't use model=TRUE
rpartpar.nosavemodel <- decision_tree(mode = "regression", min_n=1, cost_complexity=.0001) %>%
set_engine("rpart") %>%
fit(resp~num+fac+int+date+str, data = training(splits))
cat("\nrpart.rules(rpartpar.nosavemodel$fit)\n")
options(warn=2)
expect.err(try(rpart.rules(rpartpar.nosavemodel$fit)),
"Cannot retrieve the data used to build the model")
options(warn=1)
expect.err(try(plotmo(rpartpar.nosavemodel)),
"Cannot plot parsnip rpart model: need model=TRUE in call to rpart")
rpart.sqrtnum <- decision_tree(mode = "regression", min_n=1, cost_complexity=.0001) %>%
set_engine("rpart", model=TRUE) %>%
fit(resp~sqrt(num)+fac+int+date+ord+str, data = training(splits))
cat("\nrpart.rules(rpart.sqrtnum$fit)\n")
print(rpart.rules(rpart.sqrtnum$fit)) # ok
#$$ TODO
# expect.err(try(plotmo(rpart.sqrtnum)),
# "cannot get the original model predictors")
#-----------------------------------------------------------------------------------
# Test fix for github bug report https://github.com/tidymodels/parsnip/issues/341
# (fixed Sep 2020)
cat("===m750a first example===\n")
set.seed(2020)
m750a <- m4_monthly %>%
filter(id == "M750") %>%
select(-id)
print(m750a) # a tibble
set.seed(2020)
splits_a <- initial_time_split(m750a, prop = 0.9)
earth_m750a <- earth(log(value) ~ as.numeric(date) + month(date, label = TRUE), data = training(splits_a), degree=2)
print(summary(earth_m750a))
set.seed(2020)
model_m750a <- mars(mode = "regression", prod_degree=2) %>%
set_engine("earth") %>%
fit(log(value) ~ as.numeric(date) + month(date, label = TRUE), data = training(splits_a))
print(summary(model_m750a$fit))
stopifnot(identical(earth_m750a$coeff, model_m750a$fit$coeff))
predict_earth_m750a <- predict(earth_m750a, newdata=testing(splits_a)[1:3,])
predict_m750a <- model_m750a %>% predict(testing(splits_a)[1:3,])
stopifnot(max(c(9.238049628, 9.240535151, 9.232361834) - predict_m750a) < 1e-8)
stopifnot(max(predict_earth_m750a - predict_m750a) < 1e-20)
par(mfrow = c(2, 2), mar = c(3, 3, 3, 1), mgp = c(1.5, 0.5, 0))
set.seed(2020)
plotmo(model_m750a, trace=2, do.par=FALSE, pt.col="green", main="model_m750a", SHOWCALL=TRUE)
set.seed(2020)
plotmo(model_m750a$fit, trace=1, do.par=FALSE, pt.col="green", main="model_m750a$fit", SHOWCALL=TRUE)
set.seed(2020)
plotmo(earth_m750a, trace=1, do.par=FALSE, pt.col="green", main="earth_m750a", SHOWCALL=TRUE)
par(org.par)
cat("===m750a second example===\n")
set.seed(2020)
m750b <- m4_monthly %>%
filter(id == "M750") %>%
select(-id) %>%
rename(date2 = date)
print(m750b) # tibble
set.seed(2020)
splits_b <- initial_time_split(m750b, prop = 0.9)
set.seed(2020)
model_m750b <- mars(mode = "regression") %>%
set_engine("earth") %>%
fit(log(value) ~ as.numeric(date2) + month(date2, label = TRUE), data = training(splits_b))
# new data that only contains the feature "date" as a predictor
future_data <- m750b %>% future_frame(date2, .length_out = "3 years")
print(future_data) # a tibble with a single column of class "Date"
stopifnot(class(future_data[,1,drop=TRUE]) == "Date")
predict_m750a <- model_m750b %>% predict(new_data = future_data)
par(mfrow = c(2, 2), mar = c(3, 3, 3, 1), mgp = c(1.5, 0.5, 0))
set.seed(2020)
plotmo(model_m750b, trace=2, do.par=FALSE, pt.col="green", main="model_m750b", SHOWCALL=TRUE)
set.seed(2020)
plotmo(model_m750b$fit, trace=1, do.par=FALSE, pt.col="green", main="model_m750b$fit", SHOWCALL=TRUE)
par(org.par)
#-----------------------------------------------------------------------------------
# multiple response earth model
data(etitanic)
etit <- etitanic
etit$survived <- factor(ifelse(etitanic$survived == 1, "yes", "no"),
levels = c("yes", "no"))
etit$notsurvived <- factor(ifelse(etitanic$survived == 0, "notsurvived", "survived"),
levels = c("notsurvived", "survived"))
set.seed(2020)
earth_tworesp <- earth(survived + notsurvived ~ ., data=etit, degree=2)
print(summary(earth_tworesp))
# TODO following commented out because parsnip (version 0.1.5) says "'+' not meaningful for factors"
# set.seed(2020)
# mars_tworesp <- mars(mode = "regression", prod_degree=2) %>%
# set_engine("earth") %>%
# fit(survived + notsurvived~., data=etit)
# print(summary(mars_tworesp))
# print(summary(mars_tworesp$fit))
#
# stopifnot(identical(earth_tworesp$coeff, mars_tworesp$fit$coeff))
#
# predict.earth_tworesp <- predict(earth_tworesp, etit[3:6,])
# predict.mars_tworesp <- mars_tworesp %>% predict(etit[3:6,])
# stopifnot(all(predict.earth_tworesp == predict.mars_tworesp))
#
# plotmo(earth_tworesp, trace=0, nresponse=1, SHOWCALL=TRUE)
# plotmo(mars_tworesp, trace=0, nresponse=1, SHOWCALL=TRUE)
# plotmo(mars_tworesp, trace=0, nresponse=2, SHOWCALL=TRUE)
source("test.epilog.R")
|
