-
Notifications
You must be signed in to change notification settings - Fork 0
/
10. AprilTest.R
718 lines (622 loc) · 41.4 KB
/
10. AprilTest.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
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
### KTH, Department of Industrial Information and Control Systems (ICS)
### Master Thesis Project
### BUILDING PREDICTIVE MODELS FOR DYNAMIC LINE RATING USING DATA SCIENCE TECHNIQUES
### Sponsored by Vattenfall AG
### Author: Nicolae Doban
### Stockholm, 2015
## This module builds the hourly mean values and builds the predictive models on the historical observations
## Packages used: base, histogram, reshape, zoo, splines, caret
## INPUT DATA: The minute-based dataset of April, 2015
## OUTPUT: Benchmarking the predictive models for Current, CondTemp and NormalDLR
##The script-block from 16th to 200th line is similar to "1. Processing.R"- it processes and uses similar
##logic conditions on the April dataset. Later until the 242nd line, the hourly mean values of the April dataset
##are built. The following script lines runs the models (which were built on training dataset) on the test
##dataset i.e. April dataset
library(histogram)
setwd("C:/Users/LabStudent/Desktop/Nick/GitHub")
April <- read.csv("Datafile_Doban_3April.csv", header=TRUE, na.strings=c("No Data"))
NrOFDays.April<-30
April <- head(April, NrOFDays.April*24*60)
#Deleting the unnecessary columns
April$CTM_DLR=April$X.1=April$Donut2_Tension=April$X=
April$Donut2_Sag=April$Donut2_Inclinometer=
April$Donut2_Low_Point=April$X.2=
April$ZL8WSHumidity=April$ZL8WSBarometricPressure<-NULL
#/Deleting the unnecessary columns
histogram(April$NormalDLR)
#Build the temporal variables
April$Date<-as.Date(April$Date, "%m/%d/%Y")
April$Time<-as.character(April$Time)
April$DateTime<-paste(April$Date, April$Time, sep=" ")
April$DateTime<-strptime(April$DateTime, "%Y-%m-%d %H:%M")
April$Hours<-strftime(April$DateTime, "%H")
#April$Time<-as.character(April$Time)
April$Month<-months(April$Date)
#Create and re-arrange the DateTime column used for plotting versus hours
April$Days<-weekdays(as.Date(April$Date, format="%Y-%m-%d"))
#/Build the temporal variables
library(zoo)
library(splines)
interp<-data.frame()
interp<-data.frame(Date=April$Date, Time=April$Time, #SolarRad=April$SolarRad,
interpSolarRad=na.approx(April$SolarRad),
#ErrorSolarRad=abs(((na.approx(April$SolarRad)-April$SolarRad)/April$SolarRad)*100),
#WindDir=April$WindDir,
interpWindDir=na.approx(April$WindDir),
#ErrorWindDir=abs(((na.approx(April$WindDir)-April$WindDir)/April$WindDir)*100),
#AmbTemp=April$AmbTemp,
interpAmbTemp=na.spline(April$AmbTemp),
#ErrorAmbTemp=abs(((na.spline(April$AmbTemp)-April$AmbTemp)/April$AmbTemp)*100),
#WindSpeed=April$WindSpeed,
interpWindSpeed=na.approx(April$WindSpeed),
#ErrorWindSpeed=abs(((na.approx(April$WindSpeed)-April$WindSpeed)/April$WindSpeed)*100),
#CondTemp=April$CondTemp,
interpCondTemp=na.spline(April$CondTemp),
#ErrorCondTemp=abs(((na.spline(April$CondTemp)-April$CondTemp)/April$CondTemp)*100),
#Current=April$Current,
interpCurrent=na.approx(April$Current),
#ErrorCurrent=abs(((na.approx(April$Current)-April$Current)/April$Current)*100),
#NormalDLR=April$NormalDLR,
interpNormalDLR=na.spline(April$NormalDLR)
#ErrorNormalDLR=abs(((na.spline(April$NormalDLR)-April$NormalDLR)/April$NormalDLR)*100),
)
interp$Date<-as.character(interp$Date)
interp$DateTime<-paste(interp$Date, interp$Time, sep=" ")
interp$DateTime<-strptime(interp$DateTime, "%Y-%m-%d %H:%M")
interp$Hours<-strftime(interp$DateTime, "%H")
#interp$Hours<-as.character(interp$Hours)
interp$indexDT<-as.factor(as.character(interp$DateTime))
#interp$DateTime<-as.character(interp$DateTime)
interp$Days<-weekdays(as.Date(interp$Date, format="%Y-%m-%d"))
#Dealing with strange data
#Condition on WindSpeed
ptm <- proc.time()
interp$indexDT<-as.numeric(interp$indexDT)
for(i in 1:length(interp$interpWindSpeed)){
if(interp$interpWindSpeed[i]< 0.1){
interp$interpWindSpeed[i]<-interp$interpWindSpeed[i+1]+((interp$indexDT[i]-interp$indexDT[i+1])*(interp$interpWindSpeed[i-1]-interp$interpWindSpeed[i+1]))/(interp$indexDT[i-1]-interp$indexDT[i+1])
}
if(interp$interpWindSpeed[i]>10){
interp$interpWindSpeed[i]<-interp$interpWindSpeed[i+1]+((interp$indexDT[i]-interp$indexDT[i+1])*(interp$interpWindSpeed[i-1]-interp$interpWindSpeed[i+1]))/(interp$indexDT[i-1]-interp$indexDT[i+1])
}
}
proc.time() - ptm
#k<-data.frame(indexDT=interp$indexDT[interp$interpWindSpeed<0], WindSpeed=interp$interpWindSpeed[interp$interpWindSpeed<0])
#/Condition on WindSpeed
#Condition on AmbTemp
ptm <- proc.time()
interp$indexDT<-as.numeric(interp$indexDT)
for(i in 1:length(interp$interpAmbTemp)){
if(interp$interpAmbTemp[i]< -45){
interp$interpAmbTemp[i]<-interp$interpAmbTemp[i+1]+((interp$indexDT[i]-interp$indexDT[i+1])*(interp$interpAmbTemp[i-1]-interp$interpAmbTemp[i+1]))/(interp$indexDT[i-1]-interp$indexDT[i+1])
}
if(interp$interpAmbTemp[i]>40){
interp$interpAmbTemp[i]<-interp$interpAmbTemp[i+1]+((interp$indexDT[i]-interp$indexDT[i+1])*(interp$interpAmbTemp[i-1]-interp$interpAmbTemp[i+1]))/(interp$indexDT[i-1]-interp$indexDT[i+1])
}
}
proc.time() - ptm
#k<-data.frame(DateTime=interp$DateTime[interp$interpAmbTemp==NA], AmbTemp=interp$interpAmbTemp[interp$interpAmbTemp==NA])
#/Condition on AmbTemp
#Condition on SolarRad
# Start the clock!
ptm <- proc.time()
for(i in 1:length(interp$interpSolarRad)){
if(interp$interpSolarRad[i]<0.01){
interp$interpSolarRad[i]<-0
}
if(interp$interpSolarRad[i]>1400){
#some kind of alarm ?
# measuring device malfunction
#interp$interpSolarRad[i]<-0
}
}
proc.time() - ptm
#/Condition on SolarRad
#Condition on WindDir
ptm <- proc.time()
for(i in 1:length(interp$interpWindDir)){
if(interp$interpWindDir[i]<0.1){
interp$interpWindDir[i]<-0
}
if(interp$interpWindDir[i]>360){
interp$interpWindDir[i]<-360
}
}
proc.time() - ptm
#/Condition on WindDir
#Condition on CondTemp
ptm <- proc.time()
for(i in 1:length(interp$interpCondTemp)){
if(interp$interpCondTemp[i]< -20){
interp$interpCondTemp[i]<-interp$interpCondTemp[i+1]+((interp$indexDT[i]-interp$indexDT[i+1])*(interp$interpCondTemp[i-1]-interp$interpCondTemp[i+1]))/(interp$indexDT[i-1]-interp$indexDT[i+1])
}
if(interp$interpCondTemp[i]>45){
#send an signal
#interp$interpCondTemp[i]<-interp$interpCondTemp[i+1]+((interp$indexDT[i]-interp$indexDT[i+1])*(interp$interpCondTemp[i-1]-interp$interpCondTemp[i+1]))/(interp$indexDT[i-1]-interp$indexDT[i+1])
}
if(interp$interpCondTemp[i]>48){
#send an alarm
}
}
proc.time() - ptm
#/Condition on CondTemp
interp$MonthYearAbb<-strftime(as.character(interp$Date), "%b, %Y")
interp$MonthYearAbb<-as.character(interp$MonthYearAbb)
#new <- interp[interp$MonthYearAbb %in% c("Sep, 2014", "Oct, 2014", "Nov, 2014", "Dec, 2014", "Jan, 2015",
# "Feb, 2015", "Mar, 2015"), ] #English Language
interp$MonthYearAbb<-as.character(interp$MonthYearAbb)
new.April <- interp[interp$MonthYearAbb %in% c("apr, 2015"), ] #Swedish language
rownames(new.April)<-NULL
new.April$Date<-as.character(new.April$Date)
new.April$DateTime<-paste(new.April$Date, new.April$Time, sep=" ")
new.April$DateTime<-strptime(new.April$DateTime, "%Y-%m-%d %H:%M")
new.April$Hours<-strftime(new.April$DateTime, "%H")
new.April$interpCTMDLR<-NULL
new.April$indexDT<-NULL
#Replacing NAs in NormalDLR with Mean NormalDLR of that day
ptm <- proc.time()
for(i in 1:length(new.April$interpNormalDLR)){
if(is.na(new.April$interpNormalDLR[i])){
new.April$interpNormalDLR[i]<-mean(April$interpNormalDLR[April$Days[i]==April$Days[i]], na.rm=TRUE)
}
}
proc.time() - ptm
#/Replacing NAs in interpNormalDLR with Mean interpNormalDLR of that day
#Replacing NAs in Current with Mean Current of that day
ptm <- proc.time()
for(i in 1:length(new.April$interpCurrent)){
if(is.na(new.April$interpCurrent[i])){
new.April$interpCurrent[i]<-mean(April$interpCurrent[April$Days[i]==April$Days[i]], na.rm=TRUE)
}
}
proc.time() - ptm
#/Replacing NAs in Current with Mean Current of that day
#Factor the Days
new.April$Days<-as.character(new.April$Days)
#new.April$Days<-factor(new.April$Days, levels=c("Monday", "Tuesday", "Wednesday", "Thursday", "Friday",
# "Saturday", "Sunday")) #English
new.April$Days<-as.character(new.April$Days)
new.April$Days<-factor(new.April$Days, levels=c("måndag", "tisdag", "onsdag", "torsdag", "fredag",
"lördag", "söndag")) #Swedish
#Make MonthYearAbb a factor
new.April$MonthYearAbb<-as.character(new.April$MonthYearAbb)
new.April$MonthYearAbb<-factor(new.April$MonthYearAbb, levels=unique(new.April$MonthYearAbb))
#/Make MonthYearAbb a factor
#/Processing
new.April$Hours<-NULL
new.April$Days<-NULL
new.April$MonthYearAbb<-NULL
new.April$Date<-NULL
new.April$Time<-NULL
new.April$Current<-NULL
new.April<-new.April[c("DateTime", "interpSolarRad", "interpWindDir", "interpAmbTemp", "interpWindSpeed", "interpCondTemp",
"interpCurrent", "interpNormalDLR")] #Re-arrange by columns' names
library(reshape)
new.April<-rename(new.April, c("interpSolarRad"="SolarRad", "interpWindDir"="WindDir", "interpAmbTemp"="AmbTemp",
"interpWindSpeed"="WindSpeed", "interpCondTemp"="CondTemp", "interpCurrent"="Current",
"interpNormalDLR"="NormalDLR"))
NEW.April.interp<-new.April #duplicate the tidy dataframe
#Better way to average the hourly values
#because the above method does not give me a dataframe. The below method gives me a dataframe
SolarRad.Hourly.April<-aggregate(list(SolarRad = NEW.April.interp$SolarRad),
list(DateHour = cut(NEW.April.interp$DateTime, "1 hour")), mean)
WindDir.Hourly.April<-aggregate(list(WindDir = NEW.April.interp$WindDir),
list(DateHour = cut(NEW.April.interp$DateTime, "1 hour")), mean)
AmbTemp.Hourly.April<-aggregate(list(AmbTemp = NEW.April.interp$AmbTemp),
list(DateHour = cut(NEW.April.interp$DateTime, "1 hour")), mean)
WindSpeed.Hourly.April<-aggregate(list(WindSpeed = NEW.April.interp$WindSpeed),
list(DateHour = cut(NEW.April.interp$DateTime, "1 hour")), mean)
CondTemp.Hourly.April<-aggregate(list(CondTemp = NEW.April.interp$CondTemp),
list(DateHour = cut(NEW.April.interp$DateTime, "1 hour")), mean)
Current.Hourly.April<-aggregate(list(Current = NEW.April.interp$Current),
list(DateHour = cut(NEW.April.interp$DateTime, "1 hour")), mean)
NormalDLR.Hourly.April<-aggregate(list(NormalDLR = NEW.April.interp$NormalDLR),
list(DateHour = cut(NEW.April.interp$DateTime, "1 hour")), mean)
mean.Hours.April<-data.frame(DateTime=SolarRad.Hourly.April$DateHour, SolarRad=SolarRad.Hourly.April$SolarRad,
WindDir=WindDir.Hourly.April$WindDir, AmbTemp=AmbTemp.Hourly.April$AmbTemp,
WindSpeed=WindSpeed.Hourly.April$WindSpeed, CondTemp=CondTemp.Hourly.April$CondTemp,
Current=Current.Hourly.April$Current, NormalDLR=NormalDLR.Hourly.April$NormalDLR)
rownames(mean.Hours.April)<-NULL
#write.table(mean.Hours.April, "mean.Hours.April.csv", sep=",") #save to CSV file
#/Averaging the hourly dataframe from interpolated values
April.test<-mean.Hours.April
April.test$DateTime<-as.POSIXct(April.test$DateTime)
April.test$Days<-weekdays(April.test$DateTime)
April.test$Hours<-format(April.test$DateTime, "%H"); April.test$Hours<-as.numeric(April.test$Hours)
xApril <- c(rep("Weekday",24), rep("Weekday",24), rep("Weekday",24), rep("Weekend",24), rep("Weekend",24), rep("Weekday",24), rep("Weekday",24))
xW<-rep(xApril,4)
rest<-rep("Weekday",48)
April.test$daytype<-c(xW,rest)
April.test$daytype<-factor(April.test$daytype)
#re-arange columns so that (Current, CondTemp and NormalDLR) are the last ones
April.test<-April.test[c("DateTime", "SolarRad", "WindDir", "AmbTemp", "WindSpeed", "Days", "Hours", "daytype",
"Current","CondTemp", "NormalDLR")] #re-arrange columns to match the ones in the training set
April.test$DateTime<-NULL
NormalDLR.April<-April.test$NormalDLR
#April.test$NormalDLR<-NULL
#testing Current's model on April
load("C:/Users/LabStudent/Desktop/Nick/GitHub/SavedModels/PredModHistoricalObs/Current/cubist.Current.model")
April.test$pred.Current.cubist<-predict(model.Current.cubist, April.test)
April.test$err.Current.cubist<-abs(100*(April.test$Current-April.test$pred.Current.cubist)/
April.test$Current)
#plot the real versus the predicted values
plot(April.test$Current~April.test$pred.Current.cubist, main="Measured vs Predicted Current Values. April Test Dataset", xlab="Predicted Current Values [A]", ylab="Measured Current Values [A]")
April.test$DateTime<-seq.POSIXt(from=as.POSIXct("2015-04-01 00:00:00", tz="Etc/GMT+12"),
to=as.POSIXct("2015-04-30 23:00:00", tz="Etc/GMT+12"), by=3600)
#Compare the real and predicted values on a timescale
pdf("CurrentAprilTest.pdf", width=18, height=10)
plot(April.test$Current~April.test$DateTime, main="Predicted and real Current values",
xlab="Time", ylab="Current [A]", cex.lab=1.7, cex.axis=1.7)
lines(col="red", April.test$pred.Current.cubist~April.test$DateTime)
legend("topleft",
c("Measured","Predicted. cubist"), # puts text in the legend
lty=c(1,1), # gives the legend appropriate symbols (lines)
lwd=c(2.5, 2.5),col=c("black","red"),# gives the legend lines the correct color and width
cex=1.55, pt.cex=1.5, text.font=1)
dev.off()
#/Compare the real and predicted values on a timescale
#/testing Current's model on April
#CondTemp
#load models for CondTemp
load("C:/Users/LabStudent/Desktop/Nick/GitHub/SavedModels/PredModHistoricalObs/CondTemp/model.CondTemp.Formula.model")
load("C:/Users/LabStudent/Desktop/Nick/GitHub/SavedModels/PredModHistoricalObs/CondTemp/gcvEarth.CondTemp.model")
load("C:/Users/LabStudent/Desktop/Nick/GitHub/SavedModels/PredModHistoricalObs/CondTemp/brnn.CondTemp.model")
load("C:/Users/LabStudent/Desktop/Nick/GitHub/SavedModels/PredModHistoricalObs/CondTemp/bagEarth.CondTemp.model")
load("C:/Users/LabStudent/Desktop/Nick/GitHub/SavedModels/PredModHistoricalObs/CondTemp/svmLinear.CondTemp.model")
load("C:/Users/LabStudent/Desktop/Nick/GitHub/SavedModels/PredModHistoricalObs/CondTemp/bagEarthGCV.CondTemp.model")
load("C:/Users/LabStudent/Desktop/Nick/GitHub/SavedModels/PredModHistoricalObs/CondTemp/cubist.CondTemp.model")
#/load models for CondTemp
#testing CondTemp's models on April
April.test$pred.CondTemp.Formula <- predict(model.CondTemp.Formula, April.test)
April.test$err.CondTemp.Formula<-abs(100*(April.test$CondTemp-April.test$pred.CondTemp.Formula)/
April.test$CondTemp)
April.test$pred.CondTemp.cubist <- predict(model.CondTemp.cubist, April.test)
April.test$err.CondTemp.cubist<-abs(100*(April.test$CondTemp-April.test$pred.CondTemp.cubist)/
April.test$CondTemp)
April.test$pred.CondTemp.bagEarthGCV <- predict(model.CondTemp.bagEarthGCV, April.test)
April.test$err.CondTemp.bagEarthGCV<-abs(100*(April.test$CondTemp-April.test$pred.CondTemp.bagEarthGCV)/
April.test$CondTemp)
April.test$pred.CondTemp.bagEarth <- predict(model.CondTemp.bagEarth, April.test)
April.test$err.CondTemp.bagEarth<-abs(100*(April.test$CondTemp-April.test$pred.CondTemp.bagEarth)/
April.test$CondTemp)
April.test$pred.CondTemp.gcvEarth <- predict(model.CondTemp.gcvEarth, April.test)
April.test$err.CondTemp.gcvEarth<-abs(100*(April.test$CondTemp-April.test$pred.CondTemp.gcvEarth)/
April.test$CondTemp)
April.test$pred.CondTemp.svmLinear <- predict(model.CondTemp.svmLinear, April.test)
April.test$err.CondTemp.svmLinear<-abs(100*(April.test$CondTemp-April.test$pred.CondTemp.svmLinear)/
April.test$CondTemp)
April.test$pred.CondTemp.brnn <- predict(model.CondTemp.brnn, April.test)
April.test$err.CondTemp.brnn<-abs(100*(April.test$CondTemp-April.test$pred.CondTemp.brnn)/
April.test$CondTemp)
summary(April.test) #analyze the performances of the models
#/testing CondTemp's models on April
#REC curves
#REC curves represent an indicator for model performance
#First REC graph
L=vector("list",4); pred=vector("list",1); test=vector("list",1);
pred[[1]]=April.test$CondTemp;
test[[1]]=April.test$pred.CondTemp.Formula; L[[1]]=list(pred=pred,test=test,runs=1)
test[[1]]=April.test$pred.CondTemp.cubist; L[[2]]=list(pred=pred,test=test,runs=1)
test[[1]]=April.test$pred.CondTemp.bagEarthGCV; L[[3]]=list(pred=pred,test=test,runs=1)
test[[1]]=April.test$pred.CondTemp.bagEarth; L[[4]]=list(pred=pred,test=test,runs=1)
CondTemp.Formula.NAREC = mmetric(y=April.test$CondTemp, x = April.test$pred.CondTemp.Formula,
metric= "NAREC", aggregate="mean", val=5)
CondTemp.Formula.NAREC.curve = paste("REC, Formula=", round(CondTemp.Formula.NAREC, digits=5))
CondTemp.cubist.NAREC = mmetric(y=April.test$CondTemp, x = April.test$pred.CondTemp.cubist,
metric= "NAREC", aggregate="mean", val=5)
CondTemp.cubist.NAREC.curve = paste("REC, cubist=", round(CondTemp.cubist.NAREC, digits=4))
CondTemp.bagEarthGCV.NAREC = mmetric(y=April.test$CondTemp, x = April.test$pred.CondTemp.bagEarthGCV,
metric= "NAREC", aggregate="mean", val=5)
CondTemp.bagEarthGCV.NAREC.curve = paste("REC, bagEarthGCV=", round(CondTemp.bagEarthGCV.NAREC, digits=4))
CondTemp.bagEarth.NAREC = mmetric(y=April.test$CondTemp, x = April.test$pred.CondTemp.bagEarth,
metric= "NAREC", aggregate="mean", val=5)
CondTemp.bagEarth.NAREC.curve = paste("REC, bagEarth=", round(CondTemp.bagEarth.NAREC, digits=4))
#Second REC graph
B=vector("list",3); pred=vector("list",1); test=vector("list",1);
pred[[1]]=April.test$CondTemp;
test[[1]]=April.test$pred.CondTemp.gcvEarth; B[[1]]=list(pred=pred,test=test,runs=1)
test[[1]]=April.test$pred.CondTemp.svmLinear; B[[2]]=list(pred=pred,test=test,runs=1)
test[[1]]=April.test$pred.CondTemp.brnn; B[[3]]=list(pred=pred,test=test,runs=1)
CondTemp.gcvEarth.NAREC = mmetric(y=April.test$CondTemp, x = April.test$pred.CondTemp.gcvEarth,
metric= "NAREC", aggregate="mean", val=5)
CondTemp.gcvEarth.NAREC.curve = paste("REC, gcvEarth=", round(CondTemp.gcvEarth.NAREC, digits=4))
CondTemp.svmLinear.NAREC = mmetric(y=April.test$CondTemp, x = April.test$pred.CondTemp.svmLinear,
metric= "NAREC", aggregate="mean", val=5)
CondTemp.svmLinear.NAREC.curve = paste("REC, svmLinear=", round(CondTemp.svmLinear.NAREC, digits=4))
CondTemp.brnn.NAREC = mmetric(y=April.test$CondTemp, x = April.test$pred.CondTemp.brnn,
metric= "NAREC", aggregate="mean", val=5)
CondTemp.brnn.NAREC.curve = paste("REC, brnn=", round(CondTemp.brnn.NAREC, digits=4))
#save to png #png looks nicer than the pdf
png("CondTemp. REC-curves.png", width=1110, height=550)
split.screen(c(1,2))
screen(1)
L.graph<-mgraph(size=c(1,10), L, graph="REC",
leg=list(pos="bottomright",leg=c(CondTemp.gcvEarth.NAREC.curve,
CondTemp.svmLinear.NAREC.curve,
CondTemp.brnn.NAREC.curve,
CondTemp.bagEarth.NAREC.curve)),
col=c("red", "blue", "gray1", "green"),
main="CondTemp. REC curves")
screen(2)
B.graph<-mgraph(B, size=c(1,10), graph="REC",
leg=list(pos="bottomright",leg=c(CondTemp.Formula.NAREC.curve,
CondTemp.cubist.NAREC.curve,
CondTemp.bagEarthGCV.NAREC.curve)),
col=c("gray1", "blue", "green", "red"),
main="CondTemp. REC curves")
close.screen(all=TRUE)
dev.off()
#store the error metrics of the models on the test dataset
errMetric.CondTemp.brnn<-mmetric(y=April.test$CondTemp, x = April.test$pred.CondTemp.brnn,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.CondTemp.bagEarth<-mmetric(y=April.test$CondTemp, x = April.test$pred.CondTemp.bagEarth,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.CondTemp.bagEarthGCV<-mmetric(y=April.test$CondTemp, x = April.test$pred.CondTemp.bagEarthGCV,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.CondTemp.cubist<-mmetric(y=April.test$CondTemp, x = April.test$pred.CondTemp.cubist,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.CondTemp.Formula<-mmetric(y=April.test$CondTemp, x = April.test$pred.CondTemp.Formula,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.CondTemp.gcvEarth<-mmetric(y=April.test$CondTemp, x = April.test$pred.CondTemp.gcvEarth,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.CondTemp.svmLinear<-mmetric(y=April.test$CondTemp, x = April.test$pred.CondTemp.svmLinear,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMatrix.CondTemp<-data.frame(errMetric.CondTemp.bagEarth,errMetric.CondTemp.brnn,errMetric.CondTemp.cubist,
errMetric.CondTemp.Formula,errMetric.CondTemp.gcvEarth,errMetric.CondTemp.svmLinear,
errMetric.CondTemp.bagEarthGCV)
errMatrix.CondTemp<-t(errMatrix.CondTemp)
#write.table(errMatrix.CondTemp,"errorMatrixCondTemp.csv", sep=",") #optional. Save the error metric table
errMatrix.CondTemp<-errMatrix.CondTemp[,c(7,9,10,1,2,3,4,5,6,8,11)] #re-arrange the error metric table with the desired columns more to the left
#order the error metric matrix
errMatrix.CondTemp.Ordered<-errMatrix.CondTemp[ order(-errMatrix.CondTemp[,1], errMatrix.CondTemp[,2],
errMatrix.CondTemp[,3]),]
View(errMatrix.CondTemp.Ordered) #View the ordered error metric matrix.
#/REC curves
pdf("ScatterplotsCondTempAprilTest.pdf", width=18, height=10)
par(mfrow=c(3,3))
plot(April.test$CondTemp~April.test$pred.CondTemp.Formula, main="CondTemp. Measured vs Predicted by the linear model",
ylab="Measured values [C]", xlab="Predicted values [C]", cex.lab=1.7, cex.axis=1.7)
plot(April.test$CondTemp~April.test$pred.CondTemp.cubist, main="CondTemp. Measured vs Predicted by the cubist model",
ylab="Measured values [C]", xlab="Predicted values [C]", cex.lab=1.7, cex.axis=1.7)
plot(April.test$CondTemp~April.test$pred.CondTemp.bagEarthGCV, main="CondTemp. Measured vs Predicted by the bagEarthGCV model",
ylab="Measured values [C]", xlab="Predicted values [C]", cex.lab=1.7, cex.axis=1.7)
plot(April.test$CondTemp~April.test$pred.CondTemp.bagEarth, main="CondTemp. Measured vs Predicted by the bagEarth model",
ylab="Measured values [C]", xlab="Predicted values [C]", cex.lab=1.7, cex.axis=1.7)
plot(April.test$CondTemp~April.test$pred.CondTemp.brnn, main="CondTemp. Measured vs Predicted by the brnn model",
ylab="Measured values [C]", xlab="Predicted values [C]", cex.lab=1.7, cex.axis=1.7)
plot(April.test$CondTemp~April.test$pred.CondTemp.gcvEarth, main="CondTemp. Measured vs Predicted by the gcvEarth model",
ylab="Measured values [C]", xlab="Predicted values [C]", cex.lab=1.7, cex.axis=1.7)
plot(April.test$CondTemp~April.test$pred.CondTemp.svmLinear, main="CondTemp. Measured vs Predicted by the svmLinear model",
ylab="Measured values [C]", xlab="Predicted values [C]", cex.lab=1.7, cex.axis=1.7)
dev.off()
#bagEarthGCV the best model; must plot pred.CondTemp.bagEarthGCV and COndTemp against time
April.test$DateTime<-seq.POSIXt(from=as.POSIXct("2015-04-01 00:00:00", tz="Etc/GMT+12"),
to=as.POSIXct("2015-04-30 23:00:00", tz="Etc/GMT+12"), by=3600)
pdf("CondTempAprilTest.pdf", width=18, height=10)
#plot the most accurate models vs measured
plot(col="black", April.test$CondTemp~April.test$DateTime, main="Predicted and measured Conductor Temperature",
xlab="Time", ylab="Conductor Temperature [C]", cex.lab=1.7, cex.axis=1.7)
lines(col="blue", April.test$pred.CondTemp.bagEarthGCV~April.test$DateTime)
lines(col="green", April.test$pred.CondTemp.brnn~April.test$DateTime)
lines(col="red", April.test$pred.CondTemp.svmLinear~April.test$DateTime)
legend("topleft",
c("Measured","Predicted. bagEarthGCV", "Predicted. brnn", "Predicted. svmLinear"), # puts text in the legend
lty=c(1,1,1,1), # gives the legend appropriate symbols (lines)
lwd=c(2.5,2.5,2.5, 2.5),col=c("black","red", "green", "blue"),# gives the legend lines the correct color and width
cex=1.55, pt.cex=1.5, text.font=1)
#/plot the most accurate models vs measured
#/bagEarthGCV the best model; must plot pred.CondTemp.cubist and COndTemp against time
dev.off()
#bagEarthGCV the best for CondTemp
#/CondTemp
#NormalDLR
#load models for NormalDLR
load("C:/Users/LabStudent/Desktop/Nick/GitHub/SavedModels/PredModHistoricalObs/NormalDLR/cubist.NormalDLR.model")
load("C:/Users/LabStudent/Desktop/Nick/GitHub/SavedModels/PredModHistoricalObs/NormalDLR/bagEarth.NormalDLR.model")
load("C:/Users/LabStudent/Desktop/Nick/GitHub/SavedModels/PredModHistoricalObs/NormalDLR/bagEarthGCV.NormalDLR.model")
load("C:/Users/LabStudent/Desktop/Nick/GitHub/SavedModels/PredModHistoricalObs/NormalDLR/brnn.NormalDLR.model")
load("C:/Users/LabStudent/Desktop/Nick/GitHub/SavedModels/PredModHistoricalObs/NormalDLR/gcvEarth.NormalDLR.model")
load("C:/Users/LabStudent/Desktop/Nick/GitHub/SavedModels/PredModHistoricalObs/NormalDLR/model.NormalDLR.Formula.model")
load("C:/Users/LabStudent/Desktop/Nick/GitHub/SavedModels/PredModHistoricalObs/NormalDLR/svmLinear.NormalDLR.model")
#/load models for NormalDLR
#testing NormalDLR's models on April
April.test$pred.NormalDLR.Formula <- predict(model.NormalDLR.Formula, April.test)
April.test$err.NormalDLR.Formula<-abs(100*(April.test$NormalDLR-April.test$pred.NormalDLR.Formula)/
April.test$NormalDLR)
April.test$pred.NormalDLR.cubist <- predict(model.NormalDLR.cubist, April.test)
April.test$err.NormalDLR.cubist<-abs(100*(April.test$NormalDLR-April.test$pred.NormalDLR.cubist)/
April.test$NormalDLR)
April.test$pred.NormalDLR.bagEarthGCV <- predict(model.NormalDLR.bagEarthGCV, April.test)
April.test$err.NormalDLR.bagEarthGCV<-abs(100*(April.test$NormalDLR-April.test$pred.NormalDLR.bagEarthGCV)/
April.test$NormalDLR)
April.test$pred.NormalDLR.bagEarth <- predict(model.NormalDLR.bagEarth, April.test)
April.test$err.NormalDLR.bagEarth<-abs(100*(April.test$NormalDLR-April.test$pred.NormalDLR.bagEarth)/
April.test$NormalDLR)
April.test$pred.NormalDLR.gcvEarth <- predict(model.NormalDLR.gcvEarth, April.test)
April.test$err.NormalDLR.gcvEarth<-abs(100*(April.test$NormalDLR-April.test$pred.NormalDLR.gcvEarth)/
April.test$NormalDLR)
April.test$pred.NormalDLR.svmLinear <- predict(model.NormalDLR.svmLinear, April.test)
April.test$err.NormalDLR.svmLinear<-abs(100*(April.test$NormalDLR-April.test$pred.NormalDLR.svmLinear)/
April.test$NormalDLR)
April.test$pred.NormalDLR.brnn <- predict(model.NormalDLR.brnn, April.test)
April.test$err.NormalDLR.brnn<-abs(100*(April.test$NormalDLR-April.test$pred.NormalDLR.brnn)/
April.test$NormalDLR)
summary(April.test)
#/testing NormalDLR's models on April
#REC curves
#First REC graph
L=vector("list",4); pred=vector("list",1); test=vector("list",1);
pred[[1]]=April.test$NormalDLR;
test[[1]]=April.test$pred.NormalDLR.Formula; L[[1]]=list(pred=pred,test=test,runs=1)
test[[1]]=April.test$pred.NormalDLR.cubist; L[[2]]=list(pred=pred,test=test,runs=1)
test[[1]]=April.test$pred.NormalDLR.bagEarthGCV; L[[3]]=list(pred=pred,test=test,runs=1)
test[[1]]=April.test$pred.NormalDLR.bagEarth; L[[4]]=list(pred=pred,test=test,runs=1)
NormalDLR.Formula.NAREC = mmetric(y=April.test$NormalDLR, x = April.test$pred.NormalDLR.Formula,
metric= "NAREC", aggregate="mean", val=5)
NormalDLR.Formula.NAREC.curve = paste("REC, Formula=", round(NormalDLR.Formula.NAREC, digits=5))
NormalDLR.cubist.NAREC = mmetric(y=April.test$NormalDLR, x = April.test$pred.NormalDLR.cubist,
metric= "NAREC", aggregate="mean", val=5)
NormalDLR.cubist.NAREC.curve = paste("REC, cubist=", round(NormalDLR.cubist.NAREC, digits=4))
NormalDLR.bagEarthGCV.NAREC = mmetric(y=April.test$NormalDLR, x = April.test$pred.NormalDLR.bagEarthGCV,
metric= "NAREC", aggregate="mean", val=5)
NormalDLR.bagEarthGCV.NAREC.curve = paste("REC, bagEarthGCV=", round(NormalDLR.bagEarthGCV.NAREC, digits=4))
NormalDLR.bagEarth.NAREC = mmetric(y=April.test$NormalDLR, x = April.test$pred.NormalDLR.bagEarth,
metric= "NAREC", aggregate="mean", val=5)
NormalDLR.bagEarth.NAREC.curve = paste("REC, bagEarth=", round(NormalDLR.bagEarth.NAREC, digits=4))
#Second REC graph
B=vector("list",3); pred=vector("list",1); test=vector("list",1);
pred[[1]]=April.test$NormalDLR;
test[[1]]=April.test$pred.NormalDLR.gcvEarth; B[[1]]=list(pred=pred,test=test,runs=1)
test[[1]]=April.test$pred.NormalDLR.svmLinear; B[[2]]=list(pred=pred,test=test,runs=1)
test[[1]]=April.test$pred.NormalDLR.brnn; B[[3]]=list(pred=pred,test=test,runs=1)
NormalDLR.gcvEarth.NAREC = mmetric(y=April.test$NormalDLR, x = April.test$pred.NormalDLR.gcvEarth,
metric= "NAREC", aggregate="mean", val=5)
NormalDLR.gcvEarth.NAREC.curve = paste("REC, gcvEarth=", round(NormalDLR.gcvEarth.NAREC, digits=4))
NormalDLR.svmLinear.NAREC = mmetric(y=April.test$NormalDLR, x = April.test$pred.NormalDLR.svmLinear,
metric= "NAREC", aggregate="mean", val=5)
NormalDLR.svmLinear.NAREC.curve = paste("REC, svmLinear=", round(NormalDLR.svmLinear.NAREC, digits=4))
NormalDLR.brnn.NAREC = mmetric(y=April.test$NormalDLR, x = April.test$pred.NormalDLR.brnn,
metric= "NAREC", aggregate="mean", val=5)
NormalDLR.brnn.NAREC.curve = paste("REC, brnn=", round(NormalDLR.brnn.NAREC, digits=4))
#save to png #png looks nicer than the pdf
png("NormalDLR. REC-curves.png", width=1110, height=550)
split.screen(c(1,2))
screen(1)
L.graph<-mgraph(size=c(1,10), L, graph="REC",
leg=list(pos="bottomright",leg=c(NormalDLR.gcvEarth.NAREC.curve,
NormalDLR.svmLinear.NAREC.curve,
NormalDLR.brnn.NAREC.curve,
NormalDLR.bagEarth.NAREC.curve)),
col=c("red", "blue", "gray1", "green"),
main="NormalDLR. REC curves")
screen(2)
B.graph<-mgraph(B, size=c(1,10), graph="REC",
leg=list(pos="bottomright",leg=c(NormalDLR.Formula.NAREC.curve,
NormalDLR.cubist.NAREC.curve,
NormalDLR.bagEarthGCV.NAREC.curve)),
col=c("gray1", "blue", "green", "red"),
main="NormalDLR. REC curves")
close.screen(all=TRUE)
dev.off()
errMetric.NormalDLR.brnn<-mmetric(y=April.test$NormalDLR, x = April.test$pred.NormalDLR.brnn,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.NormalDLR.bagEarth<-mmetric(y=April.test$NormalDLR, x = April.test$pred.NormalDLR.bagEarth,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.NormalDLR.bagEarthGCV<-mmetric(y=April.test$NormalDLR, x = April.test$pred.NormalDLR.bagEarthGCV,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.NormalDLR.cubist<-mmetric(y=April.test$NormalDLR, x = April.test$pred.NormalDLR.cubist,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.NormalDLR.Formula<-mmetric(y=April.test$NormalDLR, x = April.test$pred.NormalDLR.Formula,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.NormalDLR.gcvEarth<-mmetric(y=April.test$NormalDLR, x = April.test$pred.NormalDLR.gcvEarth,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.NormalDLR.svmLinear<-mmetric(y=April.test$NormalDLR, x = April.test$pred.NormalDLR.svmLinear,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMatrix.NormalDLR<-data.frame(errMetric.NormalDLR.bagEarth,errMetric.NormalDLR.brnn,errMetric.NormalDLR.cubist,
errMetric.NormalDLR.Formula,errMetric.NormalDLR.gcvEarth,errMetric.NormalDLR.svmLinear,
errMetric.NormalDLR.bagEarthGCV)
errMatrix.NormalDLR<-t(errMatrix.NormalDLR)
write.table(errMatrix.NormalDLR,"errorMatrixNormalDLR.csv", sep=",")
errMatrix.NormalDLR<-errMatrix.NormalDLR[,c(7,9,10,1,2,3,4,5,6,8,11)]
errMatrix.NormalDLR.Ordered<-errMatrix.NormalDLR[ order(-errMatrix.NormalDLR[,1], errMatrix.NormalDLR[,2],
errMatrix.NormalDLR[,3]),]
View(errMatrix.NormalDLR.Ordered)
#/REC curves
pdf("ScatterplotsNormalDLRAprilTest.pdf", width=18, height=10)
par(mfrow=c(3,3))
plot(April.test$NormalDLR~April.test$pred.NormalDLR.Formula, main="NormalDLR. Calculated vs Predicted by the linear model",
ylab="Calculated values [A]", xlab="Predicted values [A]", cex.lab=1.7, cex.axis=2.5, cex.main=1.79)
plot(April.test$NormalDLR~April.test$pred.NormalDLR.cubist, main="NormalDLR. Calculated vs Predicted by the cubist model",
ylab="Calculated values [A]", xlab="Predicted values [A]", cex.lab=1.7, cex.axis=2.5, cex.main=1.79)
plot(April.test$NormalDLR~April.test$pred.NormalDLR.bagEarthGCV, main="NormalDLR. Calculated vs Predicted by the bagEarthGCV model",
ylab="Calculated values [A]", xlab="Predicted values [A]", cex.lab=1.7, cex.axis=2.5, cex.main=1.79)
plot(April.test$NormalDLR~April.test$pred.NormalDLR.bagEarth, main="NormalDLR. Calculated vs Predicted by the bagEarth model",
ylab="Calculated values [A]", xlab="Predicted values [A]", cex.lab=1.7, cex.axis=2.5, cex.main=1.79)
plot(April.test$NormalDLR~April.test$pred.NormalDLR.brnn, main="NormalDLR. Calculated vs Predicted by the brnn model",
ylab="Calculated values [A]", xlab="Predicted values [A]", cex.lab=1.7, cex.axis=2.5, cex.main=1.79)
plot(April.test$NormalDLR~April.test$pred.NormalDLR.gcvEarth, main="NormalDLR. Calculated vs Predicted by the gcvEarth model",
ylab="Calculated values [A]", xlab="Predicted values [A]", cex.lab=1.7, cex.axis=2.5, cex.main=1.79)
plot(April.test$NormalDLR~April.test$pred.NormalDLR.svmLinear, main="NormalDLR. Calculated vs Predicted by the svmLinear model",
ylab="Calculated values [A]", xlab="Predicted values [A]", cex.lab=1.7, cex.axis=2.5, cex.main=1.79)
dev.off()
#bagEarthGCV the best model; must plot NormalDLR against time
April.test$DateTime<-seq.POSIXt(from=as.POSIXct("2015-04-01 00:00:00", tz="Etc/GMT+12"),
to=as.POSIXct("2015-04-30 23:00:00", tz="Etc/GMT+12"), by=3600)
#png is better than pdf!!!
png("NormalDLRAprilTest.png", width=1100, height=550)
#plot the most accurate models vs measured
plot(col="black", April.test$NormalDLR~April.test$DateTime, main="Predicted and Calculated NormalDLR",
xlab="Time", ylab="DLR [A]", cex.lab=1.3, cex.axis=1.7, type="l")
lines(April.test$pred.NormalDLR.bagEarthGCV~April.test$DateTime,col="blue")
lines(April.test$pred.NormalDLR.bagEarth~April.test$DateTime, col="green")
lines(April.test$pred.NormalDLR.brnn~April.test$DateTime, col="red")
legend("topleft",
c("Measured","Predicted. bagEarthGCV", "Predicted. bagEarth", "Predicted. brnn"), # puts text in the legend
lty=c(1,1,1,1), # gives the legend appropriate symbols (lines)
lwd=c(2.5,2.5,2.5, 2.5),col=c("black","red", "green", "blue"),# gives the legend lines the correct color and width
cex=1.55, pt.cex=1.5, text.font=1)
dev.off()
#/png is better than pdf!!!
pdf("NormalDLRAprilTest.pdf", width=18, height=10)
#plot the most accurate models vs measured
plot(col="black", April.test$NormalDLR~April.test$DateTime, main="Predicted and Calculated NormalDLR",
xlab="Time", ylab="DLR [A]", cex.lab=1.7, cex.axis=1.7)
lines(col="blue", April.test$pred.NormalDLR.bagEarth~April.test$DateTime)
lines(col="green", April.test$pred.NormalDLR.brnn~April.test$DateTime)
lines(col="red", April.test$pred.NormalDLR.gcvEarth~April.test$DateTime)
legend("topleft",
c("Measured","Predicted. gcvEarth", "Predicted. brnn", "Predicted. svmLinear"), # puts text in the legend
lty=c(1,1,1,1), # gives the legend appropriate symbols (lines)
lwd=c(2.5,2.5,2.5, 2.5),col=c("black","red", "green", "blue"),# gives the legend lines the correct color and width
cex=1.55, pt.cex=1.5, text.font=1)
dev.off()
#/plot the most accurate models vs measured
png("ErrorRateDLR.png", width=1100, height=550)
histogram( ~ err.NormalDLR.Formula + err.NormalDLR.cubist +
err.NormalDLR.bagEarthGCV + err.NormalDLR.bagEarth +
err.NormalDLR.gcvEarth + err.NormalDLR.svmLinear +
err.NormalDLR.brnn, data=April.test, cex.axis = 1.5, cexlab = 1.7, xlab="Error rate", main="Models' error rates on the test dataset")
dev.off()
#/NormalDLR
#Benchmark the predictive models according to their error rates
#This analysis can be also be easily done for other predicted parameters
par(bg="gray92")
boxplot(April.test$err.NormalDLR.Formula, April.test$err.NormalDLR.cubist, April.test$err.NormalDLR.bagEarthGCV,
April.test$err.NormalDLR.bagEarth, April.test$err.NormalDLR.gcvEarth, April.test$err.NormalDLR.svmLinear,
April.test$err.NormalDLR.brnn, cex.lab=1.2, cex.axis=1.2, #ylim=c(0,20),
names=c("LinearModel", "Cubist", "bagEarthGCV", "bagEarth", "gcvEarth", "svmLinear", "brnn"),
col=c("red","green", "orange", "blue2", "darkslateblue", "mediumvioletred", "orangered3"),
xlab="Predictive Models", ylab="Error rate profile [%]", main="The distribution of the error rate of the predictive models for NormalDLR")
#/Benchmark the predictive models according to their error rates
#The above figure can be validated simply by running "summary(April.test)" and also by looking at
#"errMatrix.NormalDLR.Ordered" by running "View(errMatrix.NormalDLR.Ordered)"
#Plotting real values against the predicted ones
png("NormalDLRvsPredicted.png", width=1100, height=550)
par(bg="gray92")
plot(April.test$NormalDLR,April.test$NormalDLR, col="black", type="l", main="NormalDLR. Real observations VS the predicted ones",
xlab="The real values of NormalDLR [A]", ylab="The predicted values of NormalDLR [A]",
cex.lab=1.5, cex.axis=1.5)
points(April.test$NormalDLR,April.test$pred.NormalDLR.bagEarthGCV, col="blue")
points(April.test$NormalDLR,April.test$pred.NormalDLR.brnn, col="green")
points(April.test$NormalDLR,April.test$pred.NormalDLR.svmLinear, col="red")
legend("topleft",
c("Real","Predicted. bagEarthGCV", "Predicted. brnn", "Predicted. svmLinear"), # puts text in the legend
lty=c(1,1,1,1), # gives the legend appropriate symbols (lines)
lwd=c(3, 3, 3, 3),col=c("black", "blue", "green", "red"),# gives the legend lines the correct color and width
cex=1.55, pt.cex=1.55, text.font=1.4)
dev.off()
#/Plotting real values against the predicted ones