-
Notifications
You must be signed in to change notification settings - Fork 4
/
Copy pathCollaborative_Exercise_Solutions.R
422 lines (258 loc) · 14.9 KB
/
Collaborative_Exercise_Solutions.R
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
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
# This the R Code File Containing the Solutions to the Collaborative
# Exercise that concludes the Introduction to R Course available at:
# https://github.com/brfitzpatrick/Intro_to_R
# Copyright (C) 2015 Ben R. Fitzpatrick.
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program. If not, see <http://www.gnu.org/licenses/>.
#
# The course author may be contacted by email at
# This exercise uses some publically available data from the Dryad Repository
# Please download a copy of 'Data appendix.xlsx' from
# <http://dx.doi.org/10.5061/dryad.r36n0>
# When using these data, please cite the original publication:
# Gibb H, Sanders NJ, Dunn RR, Watson S, Photakis M, Abril S, Andersen AN, Angulo E, Armbrecht I, Arnan X, Baccaro FB, Bishop TR, Boulay R, Castracani C, Del Toro I, Delsinne T, Diaz M, Donoso DA, Enríquez ML, Fayle TM, Feener DH, Fitzpatrick MC, Gómez C, Grasso DA, Groc S, Heterick B, Hoffmann BD, Lach L, Lattke J, Leponce M, Lessard J, Longino J, Lucky A, Majer J, Menke SB, Mezger D, Mori A, Munyai TC, Paknia O, Pearce-Duvet J, Pfeiffer M, Philpott SM, de Souza JLP, Tista M, Vasconcelos HL, Vonshak M, Parr CL (2015) Climate mediates the effects of disturbance on ant assemblage structure. Proceedings of the Royal Society B 282(1808): 20150418. http://dx.doi.org/10.1098/rspb.2015.0418
# Additionally, please cite the Dryad data package:
# Gibb H, Sanders NJ, Dunn RR, Watson S, Photakis M, Abril S, Andersen AN, Angulo E, Armbrecht I, Arnan X, Baccaro FB, Bishop TR, Boulay R, Castracani C, Del Toro I, Delsinne T, Diaz M, Donoso DA, Enríquez ML, Fayle TM, Feener DH, Fitzpatrick MC, Gómez C, Grasso DA, Groc S, Heterick B, Hoffmann BD, Lach L, Lattke J, Leponce M, Lessard J, Longino J, Lucky A, Majer J, Menke SB, Mezger D, Mori A, Munyai TC, Paknia O, Pearce-Duvet J, Pfeiffer M, Philpott SM, de Souza JLP, Tista M, Vasconcelos HL, Vonshak M, Parr CL (2015) Data from: Climate mediates the effects of disturbance on ant assemblage structure. Dryad Digital Repository. http://dx.doi.org/10.5061/dryad.r36n0
# I have removed the second column ('Source') from the original file `Data.xlsx' before converting it to the file `Ants.csv' as we don't need the full references for each data source in the data we read into R.
# Read the Data into R:
Data <- read.csv(file = '/home/ben/Intro_to_R/Capstone_Collaborative_Exercise/Data/Ants.csv')
# Examine the top six rows of all the columns:
head(Data)
# View the Column Names:
colnames(Data)
# they're all pretty self explanatory except for PIE
# PIE = Probability of Interspecific Encounter = Probability of Two randomly selected ants being of different species
# PIE is another response variable we could attempt to model with environmental, spatial and survey characteristics
# In this example I'm going to examine the correlations between the Ant Species Richness and the covariates:
# "Mean.annual.temperature"
# "Total.annual.precipitation"
# "Temperature.range"
# "Disturbance"
# "Hemisphere"
# "Continent"
# "Pitfall.days"
# "Transect.length"
# load the 'ggplot2' library
library('ggplot2')
install.packages('maps')
install.packages('mapproj')
# maps example from http://docs.ggplot2.org/current/coord_map.html
library('maps')
library('mapproj')
world <- map_data("world")
worldmap <- ggplot(world, aes(x=long, y=lat, group=group)) +
geom_polygon(fill = 'grey') +
scale_y_continuous(breaks=(-2:2) * 30) +
scale_x_continuous(breaks=(-4:4) * 45)
# adding the sampled locations from the Ants data to this plot via an annotation layer
worldmap +
theme(panel.background = element_rect(fill = 'darkblue')) +
annotate(geom = 'point', x = Data$Longitude, y = Data$Latitude, size = 2, colour = 'red')
worldmap +
theme(panel.background = element_rect(fill = 'black'), plot.background = element_rect(fill = 'black')) +
annotate(geom = 'point', x = Data$Longitude, y = Data$Latitude, size = 3, colour = 'red') +
coord_map("ortho", orientation=c(0, -74, 0))
worldmap.ants <- worldmap +
theme(panel.background = element_rect(fill = 'black'), plot.background = element_rect(fill = 'black')) +
annotate(geom = 'point', x = Data$Longitude, y = Data$Latitude, size = 3, colour = 'red')
worldmap.ants + coord_map("ortho", orientation=c(0, 135, 0))
worldmap.ants + coord_map("ortho", orientation=c(0, -70, 0))
worldmap.ants + coord_map("ortho", orientation=c(0, -35, 0))
worldmap.ants + coord_map("ortho", orientation=c(0, 0, 0))
worldmap.ants + coord_map("ortho", orientation=c(0, 35, 0))
# This takes a few minutes to run
for(i in 1:361){
ggsave(filename = paste(i,'.png',sep =''),
plot = worldmap.ants + coord_map("ortho", orientation=c(0, 0, 0)) + coord_map("ortho", orientation=c(0, i - 180, 0)))}
# Plot Species Richness against each of the potential covariates
SR.MT <- ggplot(aes(x = Mean.annual.temperature, y = Species.richness, colour = Continent), data = Data)
SR.MT + geom_point(alpha = 0.5)
SR.TP <- ggplot(aes(x = Total.annual.precipitation, y = Species.richness, colour = Continent), data = Data)
SR.MP + geom_point(alpha = 0.5)
SR.TP <- ggplot(aes(x = Total.annual.precipitation, y = Species.richness, colour = Continent), data = Data)
SR.MP + geom_point(alpha = 0.5)
SR.MAT <- ggplot(aes(x = Mean.annual.temperature, y = Species.richness), data = Data)
SR.MAT + geom_point(alpha = 0.5)
SR.TR <- ggplot(aes(x = Temperature.range, y = Species.richness), data = Data)
SR.TR + geom_point(alpha = 0.5)
SR.D <- ggplot(aes(x = Disturbance, y = Species.richness), data = Data)
SR.D + geom_boxplot()
SR.TAP <- ggplot(aes(x = Total.annual.precipitation, y = Species.richness), data = Data)
SR.TAP + geom_point(alpha = 0.5)
SR.H <- ggplot(aes(x = Hemisphere, y = Species.richness), data = Data)
SR.H + geom_boxplot()
SR.C <- ggplot(aes(x = Continent, y = Species.richness), data = Data)
SR.C + geom_boxplot()
SR.PD <- ggplot(aes(x = Pitfall.days, y = Species.richness), data = Data)
SR.PD + geom_point(alpha = 0.5)
SR.TL <- ggplot(aes(x = Transect.length, y = Species.richness), data = Data)
SR.TL + geom_point(alpha = 0.5)
# It is good practise to recenter and rescale all the continuous covariates to the same mean and magnitude
# This is especially important as we are going to be using polynomial and interaction terms
Continuous.Covariates <- Data[ , c('Mean.annual.temperature', 'Total.annual.precipitation', 'Temperature.range', 'Pitfall.days', 'Transect.length')]
summary(Continuous.Covariates)
CC.RCRS <- scale(x = Continuous.Covariates, center = TRUE, scale = TRUE)
summary(CC.RCRS)
colMeans(CC.RCRS^2)
summary(C.RCRS)
cor(CC.RCRS)
image(abs(cor(CC.RCRS)))
colnames(CC.RCRS)
# The covariate names are a bit long so to save time let's abreivate them
# 'MAT' = 'Mean.annual.temperature'
# 'TAP' = 'Total.annual.precipitation'
# 'TR' = 'Temperature.range'
# 'PD' = 'Pitfall.days'
# 'TL' = 'Transect.length'
data.frame(Full.Name = colnames(CC.RCRS), Abbrv = c('MAT', 'TAP', 'TR', 'PD', 'TL'))
colnames(CC.RCRS) <- c('MAT', 'TAP', 'TR', 'PD', 'TL')
class(CC.RCRS)
# Creating a dataframe with polynomial effects up to order 3 for each continuous covariate term
CC.RCRS <- data.frame(CC.RCRS)
summary(CC.RCRS)
CC.RCRS.PE <- data.frame(CC.RCRS,
MAT.2 = CC.RCRS$MAT^2 ,
MAT.3 = CC.RCRS$MAT^3 ,
TAP.2 = CC.RCRS$TAP^2 ,
TAP.3 = CC.RCRS$TAP^3 ,
TR.2 = CC.RCRS$TR^2 ,
TR.3 = CC.RCRS$TR^3 ,
PD.2 = CC.RCRS$PD^2 ,
PD.3 = CC.RCRS$PD^3 ,
TL.2 = CC.RCRS$TL^2 ,
TL.3 = CC.RCRS$TL^3)
# Adding on the categorical covariates
C.RCRS <- data.frame(CC.RCRS.PE, Data[, c('Disturbance', 'Hemisphere', 'Continent')])
colnames(C.RCRS)
colnames(C.RCRS)[(ncol(C.RCRS) - 2) : ncol(C.RCRS)]
# Renaming categorical covariates to have more abreviated names
colnames(C.RCRS)[(ncol(C.RCRS) - 2) : ncol(C.RCRS)] <- c('Dist', 'Hemi', 'Cont')
colnames(C.RCRS)
# Examining a summary of our new dataframe
summary(C.RCRS)
summary(Data$Species.richness)
## Next to create some interaction terms
# for simiplicity we'll create the interaction terms for the continuous covariates first:
colnames(C.RCRS)
CC.RCRS.Lin <- C.RCRS[,c('MAT', 'TAP', 'TR', 'PD', 'TL')]
head(CC.RCRS.Lin)
dim(CC.RCRS.Lin)
choose(n = ncol(CC.RCRS.Lin), k = 2)
Int.Ind <- combn(x = 1:ncol(CC.RCRS.Lin), m = 2)
Int.Ind
Int <- data.frame(matrix(data = NA, nrow = nrow(C.RCRS), ncol = choose(n = ncol(CC.RCRS.Lin), k = 2)))
for(i in 1:ncol(Int.Ind)){
Var1 = colnames(CC.RCRS.Lin)[Int.Ind[1,i]]
Var2 = colnames(CC.RCRS.Lin)[Int.Ind[2,i]]
Int[,i] <- CC.RCRS.Lin[, Var1] * CC.RCRS.Lin[, Var2]
colnames(Int)[i] <- paste(Var1, Var2, sep = '.')
}
head(Int)
C.RCRS.Int <- data.frame(C.RCRS, Int)
head(C.RCRS.Int)
# I can't think of how to avoid doing Cont:Discrete Interactions by hand unfortunately...
# Let's set up for a stepwise variable selection
# We need to bound the search of possible model with and 'empty' intercept only model
# and a 'full' model that contains all the covariate terms
nrow(C.RCRS.Int)
# as we have 1128 observations we have sufficient degrees of freedom to fit a full model here
# this may not be the case with other data...
Empty.lm <- lm(Data$Species.richness ~ +1 , data = C.RCRS.Int)
Full.lm <- lm(Data$Species.richness ~ . + Dist*Hemi + Dist*Cont + Hemi*Cont + Dist*MAT + Dist*TAP + Dist*TR + Dist*PD + Dist*TL + Hemi*MAT + Hemi*TAP + Hemi*TR + Hemi*PD + Hemi*TL + Cont*MAT + Cont*TAP + Cont*TR + Cont*PD + Cont*TL, data = C.RCRS.Int)
summary(Full.lm)
# Note we have some NA's in the estimated coefficients:
# don't worry, not that sort of singularity ;-)
C.RCRS.Int[C.RCRS.Int$Cont == 'Asia', ]
# we just don't have much data from Asia we may need to abandon the idea of having Continent interaction terms with the other covariates:
Empty1.lm <- lm(Data$Species.richness ~ +1, data = C.RCRS.Int)
Full1.lm <- lm(Data$Species.richness ~ . + Dist*Hemi + Dist*MAT + Dist*TAP + Dist*TR + Dist*PD + Dist*TL + Hemi*MAT + Hemi*TAP + Hemi*TR + Hemi*PD + Hemi*TL, data = C.RCRS.Int)
summary(Full1.lm)
S1.lm <- step(object = Empty.lm, scope = list(lower = Empty.lm, upper = Full.lm), direction = 'both')
par(mfcol = c(2,2))
plot(S1.lm)
summary(S1.lm)
# We could drop points with inordinately high leverage on the fit... this is somewhat controversial though
Empty2.lm <- lm(Data$Species.richness[-c(113,117,189,616,940,989,1075)] ~ +1 , data = C.RCRS.Int[-c(113,117,189,616,940,989,1075),])
Full2.lm <- lm(Data$Species.richness[-c(113,117,189,616,940,989,1075)] ~ . + Dist*MAT + Dist*TAP + Dist*TR + Dist*PD + Dist*TL + Hemi*MAT + Hemi*TAP + Hemi*TR + Hemi*PD + Hemi*TL, data = C.RCRS.Int[-c(113,117,189,616,940,989,1075),])
S2.lm <- step(object = Empty2.lm, scope = list(lower = Empty2.lm, upper = Full2.lm), direction = 'both')
par(mfcol = c(2,2))
plot(S2.lm)
# Droping observations with high leverage alters the number of covariates selected:
length(coef(S1.lm))
length(coef(S2.lm))
# of the selected covariates only 19 are different:
summary(names(coef(S2.lm)) %in% names(coef(S1.lm)))
###
# Examine the predictions and associated uncertainty from the model
head(predict(S2.lm, interval = 'confidence', level = 0.95))
Pred.df <- data.frame(Observed = Data$Species.richness[-c(113,117,189,616,940,989,1075)],
predict(S2.lm, interval = 'confidence', level = 0.95))
head(Pred.df)
Fit.Obs.p <- ggplot(aes(x = Observed, y = fit), data = Pred.df)
Fit.Obs.p + geom_point() + geom_abline(slope = 1, intercept = 0) + coord_equal() + labs( y = 'Predicted') + geom_ribbon(aes(ymin = lwr, ymax = upr), alpha = 0.5, data = Pred.df) + xlim(0,30) + ylim(0,30)
##############################
## ##
## ##
## Extensions: ##
## ##
## ##
## ##
##############################
# Generalized Linear Model to use a better choice of error distribution for the species richness
# Generalized Linear Mixed Effects model to incorporate random effects for Cluster
?step # also works for objects of class glm( )
?glm
?family
# how would you go about modelling PIE?
# Recall PIE is a probability and so 0 <= PIE <= 1
# how about a logit transform? this is what Gibb et al did
# you'll need to add something to PIE before logit transforming it??? :-)
#####
#
#
# Some Visualisation Extension
#
#
#####
world <- map_data("world")
class(world)
head(world)
colnames(world)[1:2] <- c('Longitude', 'Latitude')
worldmap <- ggplot(world, aes(x=Longitude, y=Latitude, group=group)) +
geom_polygon(fill = 'grey') +
scale_y_continuous(breaks=(-2:2) * 30, label = NULL) +
scale_x_continuous(breaks=(-4:4) * 45, label = NULL)
worldmap
Data2 <- Data
Data2$group <- 1 # Dummy variable to satisfy ggplot2 when it looks for 'group' in Data2 we're not actually going to use it in the plot
worldmap +
theme(panel.background = element_rect(fill = 'white')) +
geom_point(aes(x = Longitude, y = Latitude, size = Species.richness), colour = 'red', data = Data2) +
labs(size = 'Ant Species \nRichness')
ant.map <- worldmap +
theme(panel.background = element_rect(fill = 'white'), text = element_text(size = 16)) +
geom_point(aes(x = Longitude, y = Latitude, size = Species.richness), colour = 'red', data = Data2) +
labs(size = 'Ant Species \nRichness')
ant.map + coord_map("ortho", orientation=c(0, -74, 0))
ggsave(filename = '/home/ben/Intro_to_R/Capstone_Collaborative_Exercise/Capstone_Slides_Source/Images/Ant_Sp_Rich_Map.pdf')
##
library(grid)
ant.map2 <- worldmap +
labs(size = 'Ant Species \nRichness', x = NULL, y = NULL)
ants.map2 <- ant.map2 + geom_point(aes(x = Longitude, y = Latitude, size = Species.richness), colour = 'red', data = Data2) + coord_map("ortho", orientation=c(-30, 135, 0)) + theme(panel.background = element_rect(fill = 'white'), text = element_text(size = 16), axis.ticks.length = unit(0, units = 'cm') )
ggsave(filename = '/home/ben/Intro_to_R/Capstone_Collaborative_Exercise/Capstone_Slides_Source/Images/Ant_Sp_Rich_Globe_Map.pdf', plot = ants.map2)
## Producing LaTeX Tables for slides:
library(xtable)
xtable(head(Data[,2:ncol(Data)]))