-
Notifications
You must be signed in to change notification settings - Fork 0
/
24H.R
2002 lines (1752 loc) · 123 KB
/
24H.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
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
### 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 combines the past forecasted dataset of Ambient Temperature and Wind Speed with the
## "mean.Hours" dataset and finally, builds the predictive models for the +24H time-horizon
## Packages used: base, histogram, reshape, zoo, splines, caret, boot, DAAG,
## INPUT DATA: The minute-based dataset and the past forecasted ambient temperature and wind datasets (AirTempP.csv, WindP.csv)
## OUTPUT: Building, training and testing the predictive models for the +24H time-horizon
##The order of building the predictive models: Ambient Temperature (AmbTemp), WindSpeed, Current, Conductor Temperature (CondTemp), NormalDLR
##The next predictive models are built on the previous predicted parameteres except the AmbTemp and WindSpeed
##The same approach and logic is applied when building the predictive models for the +48H time-horizon, in "48H.R"
setwd("C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/")
library(caret);library(rminer);library(boot);library(forecast);library(splines);library(DAAG); library(histogram)
mean.Hours<-read.csv("mean.Hours.csv", header=TRUE, sep=",")#import Hourly data: mean
#making sure the DateTime column is in the correct format
mean.Hours$DateTime<-seq.POSIXt(from=as.POSIXct("2014-09-01 00:00:00", tz="Etc/GMT+12"),
to=as.POSIXct("2015-03-31 23:00:00",tz="Etc/GMT+12"), by=3600)
#create temporal parameters' columns
mean.Hours$Hours<-format(mean.Hours$DateTime, "%H")
mean.Hours$Days<-weekdays(mean.Hours$DateTime)
mean.Hours$Days<-factor(mean.Hours$Days, levels=c("måndag", "tisdag", "onsdag", "torsdag", "fredag",
"lördag", "söndag")) #Swedish
mean.Hours<-mean.Hours[-c(1:336), ]; rownames(mean.Hours)<-NULL #skip the mean values of NormalDLR.
mean.Hours$Hours<-as.numeric(mean.Hours$Hours)
#Type of Day column
x <- c(rep("Weekday",24), rep("Weekday",24), rep("Weekday",24), rep("Weekday",24), rep("Weekday",24), rep("Weekend",24), rep("Weekend",24))
xW<-rep(x,28)
rest<-rep("Weekday",48)
mean.Hours$daytype<-c(xW,rest)
mean.Hours$daytype<-factor(mean.Hours$daytype)
#/create temporal parameters' columns
mean.Hours<-mean.Hours[,c(1,2,3,4,5,9,10,11,7,6,8)] #re-arrange the columns
#Because the past forecasted datasets are available every six hours, we need to extract, from mean.Hours dataset the respective observations
mean.Hours.24H<-mean.Hours[mean.Hours$Hours %in% c(0,6,12,18),]; rownames(mean.Hours.24H)<-NULL
#Importing and analyzing the AirTemp forecasted dataset
setwd("C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/") #making sure we are in the correct directory
PAirTemp<-read.csv("AirTempP.csv", sep=";", header=TRUE, skip=5120)
#PAirTemp<-head(PAirTemp, 3680)
colnames(PAirTemp)[1]<-"DateTimeLEN" #Renaming AirTemp columns
colnames(PAirTemp)[2]<-"latitud"
colnames(PAirTemp)[3]<-"longitud"
colnames(PAirTemp)[4]<-"AmbTempK" #/Renaming AirTemp columns
PAirTemp$AmbTempC<-PAirTemp$AmbTempK-273.15 #Convert AirTemp from Kelvin to Celsius
PAirTemp$LEN<-rep(c(24, 30, 36, 42, 48), 796) #new column to extract the +24H (and +48H for later on)
#+24H
library(zoo)
#Strange: Oct., 25 takes an interval from 6-hour to 5-hour - it is because+
#of the DayLight Savings Time from Oct.,25th to 26th
PAirTemp24<-PAirTemp[PAirTemp$LEN %in% c(24),]; rownames(PAirTemp24)<-NULL
#DateTime column in the correct format
PAirTemp24$DateTime<-seq.POSIXt(from=as.POSIXct("2014-09-14 00:00:00", tz="Etc/GMT+12"),
to=as.POSIXct("2015-03-31 18:00:00",tz="Etc/GMT+12"), by=3600*6)
PAirTemp24$AmbTempK<-NULL #don't need the AirTemp in Kelvin
PAirTemp24<-PAirTemp24[c("DateTime", "LEN", "latitud", "longitud", "AmbTempC" )] #re-arrange the columns
PAirTemp24<-head(PAirTemp24, 792) #Extracting only the useful observations
mean.Hours.24H$SolarRad=mean.Hours.24H$WindDir<-NULL #the SolarRad and WindDir won't be used when building the predictive models
mean.Hours.24H$PDateTime24H<-PAirTemp24$DateTime #"PDateTime24H" will be used as +24H DateTime column
mean.Hours.24H$PAmbTemp24H<-PAirTemp24$AmbTempC #"PAmbTemp24H" will be used as +24H Air Temperature column
#re-arrange the columns
mean.Hours.24H<-mean.Hours.24H[c("PDateTime24H", "PAmbTemp24H", "DateTime", "AmbTemp", "WindSpeed",
"Hours", "Days", "daytype", "Current", "CondTemp", "NormalDLR")]
#Observe a linear dependency between the prognosed and measured Air Temperature observations
plot(mean.Hours.24H$PAmbTemp24H~mean.Hours.24H$AmbTemp, ylab="AmbTemp.Prognosed [C]",
xlab="AmbTemp.Measured [C]", main="Ambtemp. Prognosed vs Measured. +24H")
#Importing and analyzing the Wind forecasted dataset
#the approach applied to the AirTemp block is applied also here with small differences
setwd("C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/") #making sure we are in the correct directory
PWind<-read.csv("WindP.csv", sep=";", header=TRUE, skip=5120)
colnames(PWind)[1]<-"DateTimeLEN" #Renaming PWind columns
colnames(PWind)[2]<-"latitud"
colnames(PWind)[3]<-"longitud"
colnames(PWind)[4]<-"u10m"
colnames(PWind)[5]<-"v10m" #/Renaming PWind columns
#PWind<-head(PWind, 3720)
PWind$PWindSpeed<-sqrt(PWind$u10m^2+PWind$v10m^2) #applying formula received from SMHI
#MUST SUBSET DATA IN +24
PWind$LEN<-rep(c(24, 30, 36, 42, 48), 796)
PWind24<-PWind[PWind$LEN %in% c(24),]; PWind24<-head(PWind24, 792); rownames(PWind24)<-NULL #only +24H
#/MUST SUBSET DATA IN +24
#+24H
PWind24$DateTime<-seq.POSIXt(from=as.POSIXct("2014-09-14 00:00:00", tz="Etc/GMT+12"),
to=as.POSIXct("2015-03-30 18:00:00",tz="Etc/GMT+12"), by=3600*6)
PWind24$DateTimeLEN=PWind24$u10m=PWind24$v10m<-NULL
PWind24<-PWind24[c("DateTime", "LEN", "latitud", "longitud","PWindSpeed")] #, "PWindDir"
PWind24$DateTime<-as.character(PWind24$DateTime)
#/Wind
#Combining the AirTemp, Wind and mean.Hours datasets
mean.Hours.24H$PWindSpeed24H<-PWind24$PWindSpeed
mean.Hours.24H<-mean.Hours.24H[c("PDateTime24H", "PAmbTemp24H", "DateTime", "AmbTemp", "PWindSpeed24H",
"WindSpeed","Hours", "Days", "daytype", "Current", "CondTemp",
"NormalDLR")]
#mean.Hours.24H$PWindSpeed24H<-mean.Hours.24H$PWindSpeed24H
#chaotic dependency between the forecasted and observed WindSpeed data
plot(mean.Hours.24H$WindSpeed~mean.Hours.24H$PWindSpeed24H, main="Wind Speed. Prognosed vs Measured. 24H",
xlab="WindSpeed.Prognosed [m/s]", ylab="WindSpeed.Measured [m/s]")
#Defining training and test datasets
train.mean.Hours.24H<-head(mean.Hours.24H, 668); rownames(train.mean.Hours.24H)<-NULL
test.mean.Hours.24H<-mean.Hours.24H[c(669:792),]; rownames(test.mean.Hours.24H)<-NULL
#building the predictive model for AirTemp. Linear Regression
model.fitAmbTemp24.lm<-lm(AmbTemp~PAmbTemp24H+I(PAmbTemp24H^2)+I(PAmbTemp24H^3),
data=train.mean.Hours.24H)
#saving the model
save(model.fitAmbTemp24.lm, file=paste('C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/24H/SavedModels/',
'model.fitAmbTemp24.lm.model',sep=''))
#Analyze the errors of the model on the train and test datasets
train.mean.Hours.24H$pred.AmbTemp.lm<-predict(model.fitAmbTemp24.lm, train.mean.Hours.24H)
train.mean.Hours.24H$err.AmbTemp<-abs(100*(train.mean.Hours.24H$AmbTemp-train.mean.Hours.24H$pred.AmbTemp.lm)/
train.mean.Hours.24H$AmbTemp)
test.mean.Hours.24H$pred.AmbTemp.lm<-predict(model.fitAmbTemp24.lm, test.mean.Hours.24H)
test.mean.Hours.24H$err.AmbTemp<-abs(100*(test.mean.Hours.24H$AmbTemp-test.mean.Hours.24H$pred.AmbTemp.lm)/
test.mean.Hours.24H$AmbTemp)
#Analyze the matching performance of the model
plot(test.mean.Hours.24H$AmbTemp~test.mean.Hours.24H$pred.AmbTemp, main="AmbTemp. Measured vs Predicted",
ylab="Measured AmbTemp [C]", xlab="Predicted AmbTemp [C]")
#plot the predicted and observed AirTemp dataset on the test dataset versus Time. Save in png file
setwd("C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/24H/SavedModels")
png("AmbTempTest.png", width=1300, height=650)
plot(test.mean.Hours.24H$AmbTemp~test.mean.Hours.24H$DateTime, main="AmbTemp. Measured and Predicted",
ylab="AmbTemp [C]", xlab="Time", cex.axis=1.5, cex.lab=1.7, type="l", col="green")
lines(test.mean.Hours.24H$pred.AmbTemp.lm~test.mean.Hours.24H$DateTime, col="red")
legend("topleft",
c("Measured","Predicted"), # puts text in the legend
lty=c(1,1), # gives the legend appropriate symbols (lines)
lwd=c(2.5, 2.5),col=c("green", "red"),# gives the legend lines the correct color and width
cex=1.55, pt.cex=1.5, text.font=1)
dev.off()
#/plot the predicted and observed AirTemp dataset on the test dataset versus Time. Save in png file
#Because the dependency between the forecasted and predicted WindSpeed data is chaotic, we must built predictive models for WindSpeed
#"train" will be used ONLY for training the data
#"train.mean.Hours.24H" will be used for storing the predicted (pred.) values of WindSpeeds by the models and their errors
train<-train.mean.Hours.24H[,c("PDateTime24H", "PWindSpeed24H", "Hours", "Days", "daytype", "WindSpeed")]
#Observe a chaotic dependency between the forecasted and measured WindSpeed data
plot(train.mean.Hours.24H$PWindSpeed24H~train.mean.Hours.24H$WindSpeed, main="Wind Speed. Train DataSet. Prognosed vs Measured. 24H",
xlab="WindSpeed.Prognosed [m/s]", ylab="WindSpeed.Measured [m/s]")
#Predicitve models for WindSpeed
#The comments found in the first predictive model's block (Linear Regression) are applicable also to the latter ones
#Linear Regression
model.fitWindSpeed24.lm<-lm(WindSpeed~poly(PWindSpeed24H,2)+Hours+Days+daytype,
data=train)
#Store the predicted values and their errors
train.mean.Hours.24H$predWindSpeed.lm<-predict(model.fitWindSpeed24.lm, train.mean.Hours.24H)
train.mean.Hours.24H$err.WindSpeed.lm<-abs(100*(train.mean.Hours.24H$WindSpeed-train.mean.Hours.24H$predWindSpeed.lm)/
train.mean.Hours.24H$WindSpeed)
#Analyze the matching performance of the model
plot(train.mean.Hours.24H$WindSpeed~train.mean.Hours.24H$predWindSpeed.lm,
main="WindSpeed. Predicted vs Measured. Linear Regression", xlab="Predicted WindSpeed [m/s]", ylab="Measured WindSpeed [m/s]")
#Analyze the summary of the predictions and of the model
summary(train.mean.Hours.24H); summary(model.fitWindSpeed24.lm)
#bagEarthGCV
train_control.bagEarthGCV <- trainControl(method="repeatedcv", number=10, repeats=3, p = 0.75)
# train the model
bagEarthGCV.Grid<-expand.grid(degree=100)
time.WindSpeed.bagEarthGCV <- proc.time()
model.WindSpeed.bagEarthGCV <- train(WindSpeed~.,
data=train, trControl=train_control.bagEarthGCV,
method="bagEarthGCV", tuneGrid = bagEarthGCV.Grid, tuneLength = 8) #tuneLength does not matter
proc.time() - time.WindSpeed.bagEarthGCV
train.mean.Hours.24H$pred.WindSpeed.bagEarthGCV<-predict(model.WindSpeed.bagEarthGCV, train[,1:5])
#record the errors
train.mean.Hours.24H$err.WindSpeed.bagEarthGCV<-abs(100*(train.mean.Hours.24H$WindSpeed-train.mean.Hours.24H$pred.WindSpeed.bagEarthGCV)/
train.mean.Hours.24H$WindSpeed)
summary(train.mean.Hours.24H)
#histogram(train.mean.Hours.24H$err.WindSpeed.bagEarthGCV)
#show error metric
errMetric.bagEarthGCV<-mmetric(y=train.mean.Hours.24H$WindSpeed, x = train.mean.Hours.24H$pred.WindSpeed.bagEarthGCV,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
print(errMetric.bagEarthGCV)
#/show error metric
#plot REC curve
mgraph(y=train.mean.Hours.24H$WindSpeed, x = train.mean.Hours.24H$pred.WindSpeed.bagEarthGCV, graph="REC",
col = "red")
#/plot REC curve
plot(train.mean.Hours.24H$WindSpeed~train.mean.Hours.24H$pred.WindSpeed.bagEarthGCV,
main="WindSpeed. Predicted vs Measured. bagEarthGCV", xlab="Predicted WindSpeed [m/s]", ylab="Measured WindSpeed [m/s]")
#/bagEarthGCV
#svmLinear
train_control.svmLinear <- trainControl(method="repeatedcv", number=10, repeats=3, p = 0.75)
# train the model
time.WindSpeed.svmLinear <- proc.time()
model.WindSpeed.svmLinear <- train(WindSpeed~.,
data=train,trControl=train_control.svmLinear,
method="svmLinear", tuneLength = 8) #tuneLength does not matter
proc.time() - time.WindSpeed.svmLinear
train.mean.Hours.24H$pred.WindSpeed.svmLinear<-predict(model.WindSpeed.svmLinear, train[,1:5])
#record the errors
train.mean.Hours.24H$err.WindSpeed.svmLinear<-abs(100*(train.mean.Hours.24H$WindSpeed-train.mean.Hours.24H$pred.WindSpeed.svmLinear)/
train.mean.Hours.24H$WindSpeed)
summary(train.mean.Hours.24H)
#histogram(train.mean.Hours.24H$err.WindSpeed.svmLinear)
#show error metric
errMetric.svmLinear<-mmetric(y=train.mean.Hours.24H$WindSpeed, x = train.mean.Hours.24H$pred.WindSpeed.svmLinear,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
print(errMetric.svmLinear)
#/show error metric
#plot REC curve
mgraph(y=train.mean.Hours.24H$WindSpeed, x = train.mean.Hours.24H$pred.WindSpeed.svmLinear, graph="REC",
col = "red")
#/plot REC curve
plot(train.mean.Hours.24H$WindSpeed~train.mean.Hours.24H$pred.WindSpeed.svmLinear,
main="WindSpeed. Predicted vs Measured. svmLinear", xlab="Predicted WindSpeed [m/s]", ylab="Measured WindSpeed [m/s]")
#/svmLinear
#gcvEarth
train_control.gcvEarth <- trainControl(method="repeatedcv", number=10, repeats=3, p = 0.75)
# train the model
gcvEarth.Grid<-expand.grid(degree=10)
time.WindSpeed.gcvEarth <- proc.time()
model.WindSpeed.gcvEarth <- train(WindSpeed~.,
data=train, trControl=train_control.gcvEarth,
method="gcvEarth", tuneLength = 8, tuneGrid=gcvEarth.Grid) #tuneLength does not matter
proc.time() - time.WindSpeed.gcvEarth
train.mean.Hours.24H$pred.WindSpeed.gcvEarth<-predict(model.WindSpeed.gcvEarth, train[,1:5])
#record the errors
train.mean.Hours.24H$err.WindSpeed.gcvEarth<-abs(100*(train.mean.Hours.24H$WindSpeed-train.mean.Hours.24H$pred.WindSpeed.gcvEarth)/
train.mean.Hours.24H$WindSpeed)
#show error metric
errMetric.gcvEarth<-mmetric(y=train.mean.Hours.24H$WindSpeed, x = train.mean.Hours.24H$pred.WindSpeed.gcvEarth,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
print(errMetric.gcvEarth)
#/show error metric
#plot REC curve
mgraph(y=train.mean.Hours.24H$WindSpeed, x = train.mean.Hours.24H$pred.WindSpeed.gcvEarth, graph="REC",
col = "red")
#/plot REC curve
plot(train.mean.Hours.24H$WindSpeed~train.mean.Hours.24H$pred.WindSpeed.gcvEarth,
main="WindSpeed. Predicted vs Measured. gcvEarth", xlab="Predicted WindSpeed [m/s]", ylab="Measured WindSpeed [m/s]")
summary(train.mean.Hours.24H)
#/gcvEarth
#cubist
train_control.cubist <- trainControl(method="repeatedcv", number=10, repeats=3, p = 0.75)
# train the model
cubist.Grid<-expand.grid(committees = 1, neighbors = 1) #highest accuracy for committees = 1, neighbors = 1
time.WindSpeed.cubist <- proc.time()
model.WindSpeed.cubist <- train(WindSpeed~.,
data=train, trControl=train_control.cubist,
method="cubist", tuneLength = 8, tuneGrid=cubist.Grid) #tuneLength does not matter
proc.time() - time.WindSpeed.cubist
train.mean.Hours.24H$pred.WindSpeed.cubist<-predict(model.WindSpeed.cubist, train[,1:5])
#record the errors
train.mean.Hours.24H$err.WindSpeed.cubist<-abs(100*(train.mean.Hours.24H$WindSpeed-train.mean.Hours.24H$pred.WindSpeed.cubist)/
train.mean.Hours.24H$WindSpeed)
#show error metric
errMetric.cubist<-mmetric(y=train.mean.Hours.24H$WindSpeed, x = train.mean.Hours.24H$pred.WindSpeed.cubist,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
print(errMetric.cubist)
#/show error metric
#plot REC curve
mgraph(y=train.mean.Hours.24H$WindSpeed, x = train.mean.Hours.24H$pred.WindSpeed.cubist, graph="REC",
col = "red")
#/plot REC curve
plot(train.mean.Hours.24H$WindSpeed~train.mean.Hours.24H$pred.WindSpeed.cubist,
main="WindSpeed. Predicted vs Measured. cubist", xlab="Predicted WindSpeed [m/s]", ylab="Measured WindSpeed [m/s]")
summary(train.mean.Hours.24H)
#/cubist
#gam
train_control.gam <- trainControl(method="repeatedcv", number=10, repeats=3, p = 0.75)
# train the model
gam.Grid<-expand.grid(select = TRUE, method = "GCV.Cp")
time.WindSpeed.gam <- proc.time()
model.WindSpeed.gam <- train(WindSpeed~.,
data=train, trControl=train_control.gam,
method="gam", tuneLength = 8, tuneGrid=gam.Grid) #tuneLength does not matter
proc.time() - time.WindSpeed.gam
train.mean.Hours.24H$pred.WindSpeed.gam<-predict(model.WindSpeed.gam, train[,1:5])
#record the errors
train.mean.Hours.24H$err.WindSpeed.gam<-abs(100*(train.mean.Hours.24H$WindSpeed-train.mean.Hours.24H$pred.WindSpeed.gam)/
train.mean.Hours.24H$WindSpeed)
#show error metric
errMetric.gam<-mmetric(y=train.mean.Hours.24H$WindSpeed, x = train.mean.Hours.24H$pred.WindSpeed.gam,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
print(errMetric.gam)
#/show error metric
#plot REC curve
mgraph(y=train.mean.Hours.24H$WindSpeed, x = train.mean.Hours.24H$pred.WindSpeed.gam, graph="REC",
col = "red")
#/plot REC curve
plot(train.mean.Hours.24H$WindSpeed~train.mean.Hours.24H$pred.WindSpeed.gam,
main="WindSpeed. Predicted vs Measured. gam", xlab="Predicted WindSpeed [m/s]", ylab="Measured WindSpeed [m/s]")
summary(train.mean.Hours.24H)
#/gam
#bagEarth
train_control.bagEarth <- trainControl(method="repeatedcv", number=10, repeats=3, p = 0.75)
# train the model
bagEarth.Grid<-expand.grid(degree=10, nprune=14)
time.WindSpeed.bagEarth <- proc.time()
model.WindSpeed.bagEarth <- train(WindSpeed~.,
data=train, trControl=train_control.bagEarth,
method="bagEarth", tuneLength = 8, tuneGrid=bagEarth.Grid) #tuneLength does not matter
proc.time() - time.WindSpeed.bagEarth
train.mean.Hours.24H$pred.WindSpeed.bagEarth<-predict(model.WindSpeed.bagEarth, train[,1:5])
#record the errors
train.mean.Hours.24H$err.WindSpeed.bagEarth<-abs(100*(train.mean.Hours.24H$WindSpeed-train.mean.Hours.24H$pred.WindSpeed.bagEarth)/
train.mean.Hours.24H$WindSpeed)
#show error metric
errMetric.bagEarth<-mmetric(y=train.mean.Hours.24H$WindSpeed, x = train.mean.Hours.24H$pred.WindSpeed.bagEarth,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
print(errMetric.bagEarth)
#/show error metric
#plot REC curve
mgraph(y=train.mean.Hours.24H$WindSpeed, x = train.mean.Hours.24H$pred.WindSpeed.bagEarth, graph="REC",
col = "red")
#/plot REC curve
plot(train.mean.Hours.24H$WindSpeed~train.mean.Hours.24H$pred.WindSpeed.bagEarth,
main="WindSpeed. Predicted vs Measured. bagEarth", xlab="Predicted WindSpeed [m/s]", ylab="Measured WindSpeed [m/s]")
summary(train.mean.Hours.24H)
#/bagEarth
#brnn
train_control.brnn <- trainControl(method="repeatedcv", number=10, repeats=3, p = 0.75)
# train the model
brnn.Grid<-expand.grid(neurons = 20)
time.WindSpeed.brnn <- proc.time()
model.WindSpeed.brnn <- train(WindSpeed~.,
data=train, trControl=train_control.brnn,
method="brnn", tuneLength = 8, tuneGrid=brnn.Grid) #tuneLength does not matter
proc.time() - time.WindSpeed.brnn
train.mean.Hours.24H$pred.WindSpeed.brnn<-predict(model.WindSpeed.brnn, train[,1:5])
#record the errors
train.mean.Hours.24H$err.WindSpeed.brnn<-abs(100*(train.mean.Hours.24H$WindSpeed-train.mean.Hours.24H$pred.WindSpeed.brnn)/
train.mean.Hours.24H$WindSpeed)
#show error metric
errMetric.brnn<-mmetric(y=train.mean.Hours.24H$WindSpeed, x = train.mean.Hours.24H$pred.WindSpeed.brnn,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
print(errMetric.brnn)
#/show error metric
#plot REC curve
mgraph(y=train.mean.Hours.24H$WindSpeed, x = train.mean.Hours.24H$pred.WindSpeed.brnn, graph="REC",
col = "red")
#/plot REC curve
plot(train.mean.Hours.24H$WindSpeed~train.mean.Hours.24H$pred.WindSpeed.brnn,
main="WindSpeed. Predicted vs Measured. brnn", xlab="Predicted WindSpeed [m/s]", ylab="Measured WindSpeed [m/s]")
summary(train.mean.Hours.24H)
#/brnn
#save the models
save(model.WindSpeed.cubist, file=paste('C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/24H/SavedModels/',
'model.WindSpeed.cubist.model',sep=''))
save(model.WindSpeed.gcvEarth, file=paste('C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/24H/SavedModels/',
'model.WindSpeed.gcvEarth.model',sep=''))
save(model.WindSpeed.bagEarthGCV, file=paste('C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/24H/SavedModels/',
'model.WindSpeed.bagEarthGCV.model',sep=''))
save(model.WindSpeed.bagEarth, file=paste('C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/24H/SavedModels/',
'model.WindSpeed.bagEarth.model',sep=''))
save(model.WindSpeed.svmLinear, file=paste('C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/24H/SavedModels/',
'model.WindSpeed.svmLinear.model',sep=''))
save(model.WindSpeed.brnn, file=paste('C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/24H/SavedModels/',
'model.WindSpeed.brnn.model',sep=''))
save(model.fitWindSpeed24.lm, file=paste('C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/24H/SavedModels/',
'model.WindSpeed.lm.model',sep=''))
save(model.WindSpeed.gam, file=paste('C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/24H/SavedModels/',
'model.WindSpeed.gam.model',sep=''))
#testing the models
test.mean.Hours.24H$pred.WindSpeed.lm<-predict(model.fitWindSpeed24.lm, test.mean.Hours.24H)
test.mean.Hours.24H$err.WindSpeed.lm<-abs(100*(test.mean.Hours.24H$WindSpeed-test.mean.Hours.24H$pred.WindSpeed.lm)/
test.mean.Hours.24H$WindSpeed)
test.mean.Hours.24H$pred.WindSpeed.cubist<-predict(model.WindSpeed.cubist, test.mean.Hours.24H)
test.mean.Hours.24H$err.WindSpeed.cubist<-abs(100*(test.mean.Hours.24H$WindSpeed-test.mean.Hours.24H$pred.WindSpeed.cubist)/
test.mean.Hours.24H$WindSpeed)
test.mean.Hours.24H$pred.WindSpeed.gcvEarth<-predict(model.WindSpeed.gcvEarth, test.mean.Hours.24H)
test.mean.Hours.24H$err.WindSpeed.gcvEarth<-abs(100*(test.mean.Hours.24H$WindSpeed-test.mean.Hours.24H$pred.WindSpeed.gcvEarth)/
test.mean.Hours.24H$WindSpeed)
test.mean.Hours.24H$pred.WindSpeed.bagEarthGCV<-predict(model.WindSpeed.bagEarthGCV, test.mean.Hours.24H)
test.mean.Hours.24H$err.WindSpeed.bagEarthGCV<-abs(100*(test.mean.Hours.24H$WindSpeed-test.mean.Hours.24H$pred.WindSpeed.bagEarthGCV)/
test.mean.Hours.24H$WindSpeed)
test.mean.Hours.24H$pred.WindSpeed.bagEarth<-predict(model.WindSpeed.bagEarth, test.mean.Hours.24H)
test.mean.Hours.24H$err.WindSpeed.bagEarth<-abs(100*(test.mean.Hours.24H$WindSpeed-test.mean.Hours.24H$pred.WindSpeed.bagEarth)/
test.mean.Hours.24H$WindSpeed)
test.mean.Hours.24H$pred.WindSpeed.svmLinear<-predict(model.WindSpeed.svmLinear, test.mean.Hours.24H)
test.mean.Hours.24H$err.WindSpeed.svmLinear<-abs(100*(test.mean.Hours.24H$WindSpeed-test.mean.Hours.24H$pred.WindSpeed.svmLinear)/
test.mean.Hours.24H$WindSpeed)
test.mean.Hours.24H$pred.WindSpeed.brnn<-predict(model.WindSpeed.brnn, test.mean.Hours.24H)
test.mean.Hours.24H$err.WindSpeed.brnn<-abs(100*(test.mean.Hours.24H$WindSpeed-test.mean.Hours.24H$pred.WindSpeed.brnn)/
test.mean.Hours.24H$WindSpeed)
test.mean.Hours.24H$pred.WindSpeed.gam<-predict(model.WindSpeed.gam, test.mean.Hours.24H)
test.mean.Hours.24H$err.WindSpeed.gam<-abs(100*(test.mean.Hours.24H$WindSpeed-test.mean.Hours.24H$pred.WindSpeed.gam)/
test.mean.Hours.24H$WindSpeed)
#Run some additional accuracy analysis
#REC curves
#First REC graph
L=vector("list",4); pred=vector("list",1); test=vector("list",1);
pred[[1]]=test.mean.Hours.24H$WindSpeed;
test[[1]]=test.mean.Hours.24H$pred.WindSpeed.lm; L[[1]]=list(pred=pred,test=test,runs=1)
test[[1]]=test.mean.Hours.24H$pred.WindSpeed.cubist; L[[2]]=list(pred=pred,test=test,runs=1)
test[[1]]=test.mean.Hours.24H$pred.WindSpeed.bagEarthGCV; L[[3]]=list(pred=pred,test=test,runs=1)
test[[1]]=test.mean.Hours.24H$pred.WindSpeed.bagEarth; L[[4]]=list(pred=pred,test=test,runs=1)
WindSpeed.lm.NAREC = mmetric(y=test.mean.Hours.24H$WindSpeed, x = test.mean.Hours.24H$pred.WindSpeed.lm,
metric= "NAREC", aggregate="mean", val=5)
WindSpeed.lm.NAREC.curve = paste("REC, Linear model=", round(WindSpeed.lm.NAREC, digits=4))
WindSpeed.cubist.NAREC = mmetric(y=test.mean.Hours.24H$WindSpeed, x = test.mean.Hours.24H$pred.WindSpeed.cubist,
metric= "NAREC", aggregate="mean", val=5)
WindSpeed.cubist.NAREC.curve = paste("REC, cubist=", round(WindSpeed.cubist.NAREC, digits=4))
WindSpeed.bagEarthGCV.NAREC = mmetric(y=test.mean.Hours.24H$WindSpeed, x = test.mean.Hours.24H$pred.WindSpeed.bagEarthGCV,
metric= "NAREC", aggregate="mean", val=5)
WindSpeed.bagEarthGCV.NAREC.curve = paste("REC, bagEarthGCV=", round(WindSpeed.bagEarthGCV.NAREC, digits=4))
WindSpeed.bagEarth.NAREC = mmetric(y=test.mean.Hours.24H$WindSpeed, x = test.mean.Hours.24H$pred.WindSpeed.bagEarth,
metric= "NAREC", aggregate="mean", val=5)
WindSpeed.bagEarth.NAREC.curve = paste("REC, bagEarth=", round(WindSpeed.bagEarth.NAREC, digits=4))
#Second REC graph
B=vector("list",4); pred=vector("list",1); test=vector("list",1);
pred[[1]]=test.mean.Hours.24H$WindSpeed;
test[[1]]=test.mean.Hours.24H$pred.WindSpeed.gcvEarth; B[[1]]=list(pred=pred,test=test,runs=1)
test[[1]]=test.mean.Hours.24H$pred.WindSpeed.svmLinear; B[[2]]=list(pred=pred,test=test,runs=1)
test[[1]]=test.mean.Hours.24H$pred.WindSpeed.brnn; B[[3]]=list(pred=pred,test=test,runs=1)
test[[1]]=test.mean.Hours.24H$pred.WindSpeed.gam; B[[4]]=list(pred=pred,test=test,runs=1)
WindSpeed.gcvEarth.NAREC = mmetric(y=test.mean.Hours.24H$WindSpeed, x = test.mean.Hours.24H$pred.WindSpeed.gcvEarth,
metric= "NAREC", aggregate="mean", val=5)
WindSpeed.gcvEarth.NAREC.curve = paste("REC, gcvEarth=", round(WindSpeed.gcvEarth.NAREC, digits=4))
WindSpeed.svmLinear.NAREC = mmetric(y=test.mean.Hours.24H$WindSpeed, x = test.mean.Hours.24H$pred.WindSpeed.svmLinear,
metric= "NAREC", aggregate="mean", val=5)
WindSpeed.svmLinear.NAREC.curve = paste("REC, svmLinear=", round(WindSpeed.svmLinear.NAREC, digits=4))
WindSpeed.brnn.NAREC = mmetric(y=test.mean.Hours.24H$WindSpeed, x = test.mean.Hours.24H$pred.WindSpeed.brnn,
metric= "NAREC", aggregate="mean", val=5)
WindSpeed.brnn.NAREC.curve = paste("REC, brnn=", round(WindSpeed.brnn.NAREC, digits=4))
WindSpeed.gam.NAREC = mmetric(y=test.mean.Hours.24H$WindSpeed, x = test.mean.Hours.24H$pred.WindSpeed.gam,
metric= "NAREC", aggregate="mean", val=5)
WindSpeed.gam.NAREC.curve = paste("REC, gam=", round(WindSpeed.gam.NAREC, digits=4))
#save to png #png looks nicer than the pdf
png("WindSpeed. REC-curves.png", width=1110, height=550)
split.screen(c(1,2))
screen(1)
L.graph<-mgraph(L, size=c(1,10), graph="REC",
leg=list(pos="bottomright",leg=c(WindSpeed.lm.NAREC.curve,
WindSpeed.cubist.NAREC.curve,
WindSpeed.bagEarthGCV.NAREC.curve,
WindSpeed.bagEarth.NAREC.curve)),
col=c("blue", "gray1", "green", "red"),
main="WindSpeed. REC curves")
screen(2)
B.graph<-mgraph(B, size=c(1,10), graph="REC",
leg=list(pos="bottomright",leg=c(WindSpeed.brnn.NAREC.curve,
WindSpeed.gcvEarth.NAREC.curve,
WindSpeed.svmLinear.NAREC.curve,
WindSpeed.gam.NAREC.curve)),
col=c("red", "gray1", "green", "blue"),
main="WindSpeed. REC curves")
close.screen(all=TRUE)
dev.off()
#Store the error metric matrices of the models
errMetric.WindSpeed.brnn<-mmetric(y=test.mean.Hours.24H$WindSpeed, x = test.mean.Hours.24H$pred.WindSpeed.brnn,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.WindSpeed.bagEarth<-mmetric(y=test.mean.Hours.24H$WindSpeed, x = test.mean.Hours.24H$pred.WindSpeed.bagEarth,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.WindSpeed.bagEarthGCV<-mmetric(y=test.mean.Hours.24H$WindSpeed, x = test.mean.Hours.24H$pred.WindSpeed.bagEarthGCV,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.WindSpeed.cubist<-mmetric(y=test.mean.Hours.24H$WindSpeed, x = test.mean.Hours.24H$pred.WindSpeed.cubist,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.WindSpeed.lm<-mmetric(y=test.mean.Hours.24H$WindSpeed, x = test.mean.Hours.24H$pred.WindSpeed.lm,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.WindSpeed.gcvEarth<-mmetric(y=test.mean.Hours.24H$WindSpeed, x = test.mean.Hours.24H$pred.WindSpeed.gcvEarth,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.WindSpeed.svmLinear<-mmetric(y=test.mean.Hours.24H$WindSpeed, x = test.mean.Hours.24H$pred.WindSpeed.svmLinear,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.WindSpeed.gam<-mmetric(y=test.mean.Hours.24H$WindSpeed, x = test.mean.Hours.24H$pred.WindSpeed.gam,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
#Combine all the metrics in one dataframe
errMatrix.WindSpeed<-data.frame(errMetric.WindSpeed.bagEarth,errMetric.WindSpeed.brnn,errMetric.WindSpeed.cubist,
errMetric.WindSpeed.lm,errMetric.WindSpeed.gcvEarth,errMetric.WindSpeed.svmLinear,
errMetric.WindSpeed.bagEarthGCV, errMetric.WindSpeed.gam)
errMatrix.WindSpeed<-t(errMatrix.WindSpeed)
#write.table(errMatrix.WindSpeed,"errorMatrixWindSpeed.csv", sep=",") #save the error metric dataframe if needed
#Re-arrange the error metric dataframe with the most important indicators to the left
errMatrix.WindSpeed<-errMatrix.WindSpeed[,c(7,9,10,1,2,3,4,5,6,8,11)]
#Order the error metric dataframe
errMatrix.WindSpeed.Ordered<-errMatrix.WindSpeed[ order(-errMatrix.WindSpeed[,1], -errMatrix.WindSpeed[,2],
errMatrix.WindSpeed[,3]),]
#View the error metric dataframe
View(errMatrix.WindSpeed.Ordered)
#/REC curves
#/testing the models
#plot the predicted and observed WindSpeed dataset on the test dataset versus Time. Save in png file
png("WindSpeedTest.png", width=1300, height=650)
plot(test.mean.Hours.24H$WindSpeed~test.mean.Hours.24H$DateTime, col="black", pch=21,
main="Wind Speed. Predicted and Measured", ylab="WindSpeed [m/s]", xlab="Time", cex.axis=1.5, cex.lab=1.7)
points(test.mean.Hours.24H$WindSpeed~test.mean.Hours.24H$DateTime, bg='black', pch=21, cex=1.5)
lines(test.mean.Hours.24H$pred.WindSpeed.lm~test.mean.Hours.24H$DateTime, col="blue")
lines(test.mean.Hours.24H$pred.WindSpeed.cubist~test.mean.Hours.24H$DateTime, col="red")
lines(test.mean.Hours.24H$pred.WindSpeed.gam~test.mean.Hours.24H$DateTime, col="green")
lines(test.mean.Hours.24H$pred.WindSpeed.svmLinear~test.mean.Hours.24H$DateTime, col="deepskyblue3")
legend("topleft",
c("Measured","Predicted. lm", "Predicted. cubist", "Predicted. gam", "Predicted. svmLinear"), # puts text in the legend
lty=c(1,1,1,1,1), # gives the legend appropriate symbols (lines)
lwd=c(2.5,2.5,2.5, 2.5, 2.5),col=c("black","blue", "red", "green", "deepskyblue3"),# gives the legend lines the correct color and width
cex=1.55, pt.cex=1.5, text.font=1)
dev.off()
#plot the predicted and observed WindSpeed dataset on the test dataset versus Time. Save in png file
#Predicted versus the measured
WindSpeed<-test.mean.Hours.24H$WindSpeed #Store the measured WindSpeed as a vector
png("WindSpeedPredictedVSMeasured.png", width=1300, height=650)
par(bg="gray85") #Changing the background color of the canvas to have a greater contrast
plot(test.mean.Hours.24H$WindSpeed~WindSpeed, col="black", pch=21, type="l",
main="Wind Speed. Predicted VS Measured", xlab="Measured WindSpeed [m/s]", ylab="Predicted WindSpeed [m/s]", cex.axis=1.5, cex.lab=1.7)
points(test.mean.Hours.24H$pred.WindSpeed.lm~WindSpeed, col="blue", cex=1.5)
points(test.mean.Hours.24H$pred.WindSpeed.cubist~WindSpeed, col="red", cex=1.5)
points(test.mean.Hours.24H$pred.WindSpeed.gam~WindSpeed, col="green", cex=1.5)
points(test.mean.Hours.24H$pred.WindSpeed.svmLinear~WindSpeed, col="yellow", cex=1.5)
legend("topleft",
c("Measured","Predicted. lm", "Predicted. cubist", "Predicted. gam", "Predicted. svmLinear"), # puts text in the legend
lty=c(1,1,1,1,1), # gives the legend appropriate symbols (lines)
lwd=c(2.5,2.5,2.5, 2.5, 2.5),col=c("black","blue", "red", "green", "yellow"),# gives the legend lines the correct color and width
cex=1.55, pt.cex=1.5, text.font=1)
dev.off()
#/Predicted versus the measured
# The WindSpeed will be predicted with Cubist model
#Work with predicted AmbTemp and WindSpeed
#Re-create the training and the test datasets with the predicted AirTemp and WindSPeed data
train.mean.Hours.24H<-head(mean.Hours.24H, 668); rownames(train.mean.Hours.24H)<-NULL
load("C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/24H/SavedModels/model.fitAmbTemp24.lm.model")
train.mean.Hours.24H$predAmbTemp.lm<-predict(model.fitAmbTemp24.lm, train.mean.Hours.24H)
load("C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/24H/SavedModels/model.WindSpeed.cubist.model")
train.mean.Hours.24H$pred.WindSpeed.cubist<-predict(model.WindSpeed.cubist, train.mean.Hours.24H)
train.mean.Hours.24H<-train.mean.Hours.24H[c("PDateTime24H", "PAmbTemp24H", "DateTime", "AmbTemp",
"predAmbTemp.lm", "PWindSpeed24H","WindSpeed",
"pred.WindSpeed.cubist","Hours", "Days", "daytype", "Current",
"CondTemp", "NormalDLR")]
train<-train.mean.Hours.24H[,c("DateTime", "predAmbTemp.lm","pred.WindSpeed.cubist", "Hours", "Days", "daytype", "Current")]
test.mean.Hours.24H<-mean.Hours.24H[c(669:792),]; rownames(test.mean.Hours.24H)<-NULL
test.mean.Hours.24H$predAmbTemp.lm<-predict(model.fitAmbTemp24.lm, test.mean.Hours.24H)
test.mean.Hours.24H$pred.WindSpeed.cubist<-predict(model.WindSpeed.cubist, test.mean.Hours.24H)
test.mean.Hours.24H<-test.mean.Hours.24H[c("PDateTime24H", "PAmbTemp24H", "DateTime", "AmbTemp", "predAmbTemp.lm",
"PWindSpeed24H","WindSpeed", "pred.WindSpeed.cubist", "Hours",
"Days", "daytype", "Current", "CondTemp", "NormalDLR")]
#Predicting Current
#Linear model
model.Current.lm=lm(Current~I(pred.WindSpeed.cubist^1)+I(predAmbTemp.lm^1)+DateTime+Days+
+I(Hours^1)*daytype*I(pred.WindSpeed.cubist^1),
data=train); accuracy(f=model.Current.lm) #MAPE
#/Linear model
train.mean.Hours.24H$pred.Current.lm<-predict(model.Current.lm, train.mean.Hours.24H)
train.mean.Hours.24H$err.Current.lm<-abs(100*(train.mean.Hours.24H$Current-train.mean.Hours.24H$pred.Current.lm)
/train.mean.Hours.24H$Current)
#svmLinear
train_control.svmLinear <- trainControl(method="repeatedcv", number=10, repeats=3, p = 0.75)
# train the model
time.Current.svmLinear <- proc.time()
model.Current.svmLinear <- train(Current~.,
data=train, trControl=train_control.svmLinear,
method="svmLinear", tuneLength = 8) #tuneLength does not matter
proc.time() - time.Current.svmLinear
train.mean.Hours.24H$pred.Current.svmLinear<-predict(model.Current.svmLinear, train.mean.Hours.24H)
#record the errors
train.mean.Hours.24H$err.Current.svmLinear<-abs(100*(train.mean.Hours.24H$Current-train.mean.Hours.24H$pred.Current.svmLinear)/
train.mean.Hours.24H$Current)
summary(train.mean.Hours.24H)
#histogram(train.mean.Hours.24H$err.Current.svmLinear)
#show error metric
errMetric.svmLinear<-mmetric(y=train.mean.Hours.24H$Current, x = train.mean.Hours.24H$pred.Current.svmLinear,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
print(errMetric.svmLinear)
#/show error metric
#plot REC curve
mgraph(y=train.mean.Hours.24H$Current, x = train.mean.Hours.24H$pred.Current.svmLinear, graph="REC",
col = "red")
#/plot REC curve
plot(train.mean.Hours.24H$Current~train.mean.Hours.24H$pred.Current.svmLinear,
main="Current. Predicted vs Measured. svmLinear", xlab="Predicted Current [A]", ylab="Measured Current [A]")
#/svmLinear
#cubist
train_control.cubist <- trainControl(method="repeatedcv", number=10, repeats=3, p = 0.75)
# train the model
cubist.Grid<-expand.grid(committees = 1, neighbors = 1) #highest accuracy for committees = 1, neighbors = 1
time.Current.cubist <- proc.time()
model.Current.cubist <- train(Current~.,
data=train, trControl=train_control.cubist,
method="cubist", tuneGrid = cubist.Grid, tuneLength = 8) #tuneLength does not matter
proc.time() - time.Current.cubist
train.mean.Hours.24H$pred.Current.cubist<-predict(model.Current.cubist, train.mean.Hours.24H)
#record the errors
train.mean.Hours.24H$err.Current.cubist<-abs(100*(train.mean.Hours.24H$Current-train.mean.Hours.24H$pred.Current.cubist)/
train.mean.Hours.24H$Current)
summary(train.mean.Hours.24H)
#histogram(train.mean.Hours.24H$err.Current.cubist)
#show error metric
errMetric.cubist<-mmetric(y=train.mean.Hours.24H$Current, x = train.mean.Hours.24H$pred.Current.cubist,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
print(errMetric.cubist)
#/show error metric
#plot REC curve
mgraph(y=train.mean.Hours.24H$Current, x = train.mean.Hours.24H$pred.Current.cubist, graph="REC",
col = "red")
#/plot REC curve
plot(train.mean.Hours.24H$Current~train.mean.Hours.24H$pred.Current.cubist,
main="Current. Predicted vs Measured. cubist", xlab="Predicted Current [A]", ylab="Measured Current [A]")
#/cubist
#bagEarthGCV
train_control.bagEarthGCV <- trainControl(method="repeatedcv", number=10, repeats=3, p = 0.75)
# train the model
bagEarthGCV.Grid<-expand.grid(degree=6)
time.Current.bagEarthGCV <- proc.time()
model.Current.bagEarthGCV <- train(Current~.,
data=train, trControl=train_control.bagEarthGCV,
method="bagEarthGCV", tuneGrid = bagEarthGCV.Grid, tuneLength = 8) #tuneLength does not matter
proc.time() - time.Current.bagEarthGCV
train.mean.Hours.24H$pred.Current.bagEarthGCV<-predict(model.Current.bagEarthGCV, train.mean.Hours.24H)
#record the errors
train.mean.Hours.24H$err.Current.bagEarthGCV<-abs(100*(train.mean.Hours.24H$Current-train.mean.Hours.24H$pred.Current.bagEarthGCV)/
train.mean.Hours.24H$Current)
summary(train.mean.Hours.24H)
#histogram(train.mean.Hours.24H$err.Current.bagEarthGCV)
#show error metric
errMetric.bagEarthGCV<-mmetric(y=train.mean.Hours.24H$Current, x = train.mean.Hours.24H$pred.Current.bagEarthGCV,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
print(errMetric.bagEarthGCV)
#/show error metric
#plot REC curve
mgraph(y=train.mean.Hours.24H$Current, x = train.mean.Hours.24H$pred.Current.bagEarthGCV, graph="REC",
col = "red")
#/plot REC curve
plot(train.mean.Hours.24H$Current~train.mean.Hours.24H$pred.Current.bagEarthGCV,
main="Current. Predicted vs Measured. bagEarthGCV", xlab="Predicted Current [A]", ylab="Measured Current [A]")
#/bagEarthGCV
#gcvEarth
train_control.gcvEarth <- trainControl(method="repeatedcv", number=10, repeats=3, p = 0.75)
# train the model
gcvEarth.Grid<-expand.grid(degree=10)
time.Current.gcvEarth <- proc.time()
model.Current.gcvEarth <- train(Current~.,
data=train, trControl=train_control.gcvEarth,
method="gcvEarth", tuneGrid = gcvEarth.Grid, tuneLength = 8) #tuneLength does not matter
proc.time() - time.Current.gcvEarth
train.mean.Hours.24H$pred.Current.gcvEarth<-predict(model.Current.gcvEarth, train.mean.Hours.24H)
#record the errors
train.mean.Hours.24H$err.Current.gcvEarth<-abs(100*(train.mean.Hours.24H$Current-train.mean.Hours.24H$pred.Current.gcvEarth)/
train.mean.Hours.24H$Current)
summary(train.mean.Hours.24H)
#histogram(train.mean.Hours.24H$err.Current.gcvEarth)
#show error metric
errMetric.gcvEarth<-mmetric(y=train.mean.Hours.24H$Current, x = train.mean.Hours.24H$pred.Current.gcvEarth,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
print(errMetric.gcvEarth)
#/show error metric
#plot REC curve
mgraph(y=train.mean.Hours.24H$Current, x = train.mean.Hours.24H$pred.Current.gcvEarth, graph="REC",
col = "red")
#/plot REC curve
plot(train.mean.Hours.24H$Current~train.mean.Hours.24H$pred.Current.gcvEarth,
main="Current. Predicted vs Measured. gcvEarth", xlab="Predicted Current [A]", ylab="Measured Current [A]")
#/gcvEarth
#gam
train_control.gam <- trainControl(method="repeatedcv", number=10, repeats=3, p = 0.75)
# train the model
gam.Grid<-expand.grid(select = TRUE, method = "GCV.Cp")
time.Current.gam <- proc.time()
model.Current.gam <- train(Current~.,
data=train, trControl=train_control.gam,
method="gam", tuneGrid = gam.Grid, tuneLength = 8) #tuneLength does not matter
proc.time() - time.Current.gam
train.mean.Hours.24H$pred.Current.gam<-predict(model.Current.gam, train.mean.Hours.24H)
#record the errors
train.mean.Hours.24H$err.Current.gam<-abs(100*(train.mean.Hours.24H$Current-train.mean.Hours.24H$pred.Current.gam)/
train.mean.Hours.24H$Current)
summary(train.mean.Hours.24H)
#histogram(train.mean.Hours.24H$err.Current.gam)
#show error metric
errMetric.gam<-mmetric(y=train.mean.Hours.24H$Current, x = train.mean.Hours.24H$pred.Current.gam,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
print(errMetric.gam)
#/show error metric
#plot REC curve
mgraph(y=train.mean.Hours.24H$Current, x = train.mean.Hours.24H$pred.Current.gam, graph="REC",
col = "red")
#/plot REC curve
plot(train.mean.Hours.24H$Current~train.mean.Hours.24H$pred.Current.gam,
main="Current. Predicted vs Measured. gam", xlab="Predicted Current [A]", ylab="Measured Current [A]")
#/gam
#bagEarth
train_control.bagEarth <- trainControl(method="repeatedcv", number=10, repeats=3, p = 0.75)
# train the model
bagEarth.Grid<-expand.grid(degree=10, nprune=14)
time.Current.bagEarth <- proc.time()
model.Current.bagEarth <- train(Current~.,
data=train, trControl=train_control.bagEarth,
method="bagEarth", tuneGrid = bagEarth.Grid, tuneLength = 8) #tuneLength does not matter
proc.time() - time.Current.bagEarth
train.mean.Hours.24H$pred.Current.bagEarth<-predict(model.Current.bagEarth, train.mean.Hours.24H)
#record the errors
train.mean.Hours.24H$err.Current.bagEarth<-abs(100*(train.mean.Hours.24H$Current-train.mean.Hours.24H$pred.Current.bagEarth)/
train.mean.Hours.24H$Current)
summary(train.mean.Hours.24H)
#histogram(train.mean.Hours.24H$err.Current.bagEarth)
#show error metric
errMetric.bagEarth<-mmetric(y=train.mean.Hours.24H$Current, x = train.mean.Hours.24H$pred.Current.bagEarth,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
print(errMetric.bagEarth)
#/show error metric
#plot REC curve
mgraph(y=train.mean.Hours.24H$Current, x = train.mean.Hours.24H$pred.Current.bagEarth, graph="REC",
col = "red")
#/plot REC curve
plot(train.mean.Hours.24H$Current~train.mean.Hours.24H$pred.Current.bagEarth,
main="Current. Predicted vs Measured. bagEarth", xlab="Predicted Current [A]", ylab="Measured Current [A]")
#/bagEarth
#brnn
train_control.brnn <- trainControl(method="repeatedcv", number=10, repeats=3, p = 0.75)
# train the model
brnn.Grid<-expand.grid(neurons = 20)
time.Current.brnn <- proc.time()
model.Current.brnn <- train(Current~.,
data=train, trControl=train_control.brnn,
method="brnn", tuneGrid = brnn.Grid, tuneLength = 8) #tuneLength does not matter
proc.time() - time.Current.brnn
train.mean.Hours.24H$pred.Current.brnn<-predict(model.Current.brnn, train.mean.Hours.24H)
#record the errors
train.mean.Hours.24H$err.Current.brnn<-abs(100*(train.mean.Hours.24H$Current-train.mean.Hours.24H$pred.Current.brnn)/
train.mean.Hours.24H$Current)
summary(train.mean.Hours.24H)
#histogram(train.mean.Hours.24H$err.Current.brnn)
#show error metric
errMetric.brnn<-mmetric(y=train.mean.Hours.24H$Current, x = train.mean.Hours.24H$pred.Current.brnn,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
print(errMetric.brnn)
#/show error metric
#plot REC curve
mgraph(y=train.mean.Hours.24H$Current, x = train.mean.Hours.24H$pred.Current.brnn, graph="REC",
col = "red")
#/plot REC curve
plot(train.mean.Hours.24H$Current~train.mean.Hours.24H$pred.Current.brnn,
main="Current. Predicted vs Measured. brnn", xlab="Predicted Current [A]", ylab="Measured Current [A]")
#/brnn
#/Predicting Current
#save the models
save(model.Current.cubist, file=paste('C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/24H/SavedModels/',
'model.Current.cubist.model',sep=''))
save(model.Current.gcvEarth, file=paste('C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/24H/SavedModels/',
'model.Current.gcvEarth.model',sep=''))
save(model.Current.bagEarthGCV, file=paste('C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/24H/SavedModels/',
'model.Current.bagEarthGCV.model',sep=''))
save(model.Current.bagEarth, file=paste('C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/24H/SavedModels/',
'model.Current.bagEarth.model',sep=''))
save(model.Current.svmLinear, file=paste('C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/24H/SavedModels/',
'model.Current.svmLinear.model',sep=''))
save(model.Current.brnn, file=paste('C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/24H/SavedModels/',
'model.Current.brnn.model',sep=''))
save(model.Current.lm, file=paste('C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/24H/SavedModels/',
'model.Current.lm.model',sep=''))
save(model.Current.gam, file=paste('C:/Users/LabStudent/Desktop/Nick/GitHub/ForecastedObservations/24H/SavedModels/',
'model.Current.gam.model',sep=''))
#/save the models
#Testing Current models
test.mean.Hours.24H$pred.Current.lm<-predict(model.Current.lm, test.mean.Hours.24H)
test.mean.Hours.24H$err.Current.lm<-abs(100*(test.mean.Hours.24H$Current-test.mean.Hours.24H$pred.Current.lm)/
test.mean.Hours.24H$Current)
test.mean.Hours.24H$pred.Current.cubist<-predict(model.Current.cubist, test.mean.Hours.24H)
test.mean.Hours.24H$err.Current.cubist<-abs(100*(test.mean.Hours.24H$Current-test.mean.Hours.24H$pred.Current.cubist)/
test.mean.Hours.24H$Current)
test.mean.Hours.24H$pred.Current.gcvEarth<-predict(model.Current.gcvEarth, test.mean.Hours.24H)
test.mean.Hours.24H$err.Current.gcvEarth<-abs(100*(test.mean.Hours.24H$Current-test.mean.Hours.24H$pred.Current.gcvEarth)/
test.mean.Hours.24H$Current)
test.mean.Hours.24H$pred.Current.bagEarthGCV<-predict(model.Current.bagEarthGCV, test.mean.Hours.24H)
test.mean.Hours.24H$err.Current.bagEarthGCV<-abs(100*(test.mean.Hours.24H$Current-test.mean.Hours.24H$pred.Current.bagEarthGCV)/
test.mean.Hours.24H$Current)
test.mean.Hours.24H$pred.Current.bagEarth<-predict(model.Current.bagEarth, test.mean.Hours.24H)
test.mean.Hours.24H$err.Current.bagEarth<-abs(100*(test.mean.Hours.24H$Current-test.mean.Hours.24H$pred.Current.bagEarth)/
test.mean.Hours.24H$Current)
test.mean.Hours.24H$pred.Current.svmLinear<-predict(model.Current.svmLinear, test.mean.Hours.24H)
test.mean.Hours.24H$err.Current.svmLinear<-abs(100*(test.mean.Hours.24H$Current-test.mean.Hours.24H$pred.Current.svmLinear)/
test.mean.Hours.24H$Current)
test.mean.Hours.24H$pred.Current.brnn<-predict(model.Current.brnn, test.mean.Hours.24H)
test.mean.Hours.24H$err.Current.brnn<-abs(100*(test.mean.Hours.24H$Current-test.mean.Hours.24H$pred.Current.brnn)/
test.mean.Hours.24H$Current)
test.mean.Hours.24H$pred.Current.gam<-predict(model.Current.gam, test.mean.Hours.24H)
test.mean.Hours.24H$err.Current.gam<-abs(100*(test.mean.Hours.24H$Current-test.mean.Hours.24H$pred.Current.gam)/
test.mean.Hours.24H$Current)
#Additional model accracy indicators
#REC curves
#First REC graph
L=vector("list",4); pred=vector("list",1); test=vector("list",1);
pred[[1]]=test.mean.Hours.24H$Current;
test[[1]]=test.mean.Hours.24H$pred.Current.lm; L[[1]]=list(pred=pred,test=test,runs=1)
test[[1]]=test.mean.Hours.24H$pred.Current.cubist; L[[2]]=list(pred=pred,test=test,runs=1)
test[[1]]=test.mean.Hours.24H$pred.Current.bagEarthGCV; L[[3]]=list(pred=pred,test=test,runs=1)
test[[1]]=test.mean.Hours.24H$pred.Current.bagEarth; L[[4]]=list(pred=pred,test=test,runs=1)
Current.lm.NAREC = mmetric(y=test.mean.Hours.24H$Current, x = test.mean.Hours.24H$pred.Current.lm,
metric= "NAREC", aggregate="mean", val=5)
Current.lm.NAREC.curve = paste("REC, Linear model=", round(Current.lm.NAREC, digits=4))
Current.cubist.NAREC = mmetric(y=test.mean.Hours.24H$Current, x = test.mean.Hours.24H$pred.Current.cubist,
metric= "NAREC", aggregate="mean", val=5)
Current.cubist.NAREC.curve = paste("REC, cubist=", round(Current.cubist.NAREC, digits=4))
Current.bagEarthGCV.NAREC = mmetric(y=test.mean.Hours.24H$Current, x = test.mean.Hours.24H$pred.Current.bagEarthGCV,
metric= "NAREC", aggregate="mean", val=5)
Current.bagEarthGCV.NAREC.curve = paste("REC, bagEarthGCV=", round(Current.bagEarthGCV.NAREC, digits=4))
Current.bagEarth.NAREC = mmetric(y=test.mean.Hours.24H$Current, x = test.mean.Hours.24H$pred.Current.bagEarth,
metric= "NAREC", aggregate="mean", val=5)
Current.bagEarth.NAREC.curve = paste("REC, bagEarth=", round(Current.bagEarth.NAREC, digits=4))
#Second REC graph
B=vector("list",4); pred=vector("list",1); test=vector("list",1);
pred[[1]]=test.mean.Hours.24H$Current;
test[[1]]=test.mean.Hours.24H$pred.Current.gcvEarth; B[[1]]=list(pred=pred,test=test,runs=1)
test[[1]]=test.mean.Hours.24H$pred.Current.svmLinear; B[[2]]=list(pred=pred,test=test,runs=1)
test[[1]]=test.mean.Hours.24H$pred.Current.brnn; B[[3]]=list(pred=pred,test=test,runs=1)
test[[1]]=test.mean.Hours.24H$pred.Current.gam; B[[4]]=list(pred=pred,test=test,runs=1)
Current.gcvEarth.NAREC = mmetric(y=test.mean.Hours.24H$Current, x = test.mean.Hours.24H$pred.Current.gcvEarth,
metric= "NAREC", aggregate="mean", val=5)
Current.gcvEarth.NAREC.curve = paste("REC, gcvEarth=", round(Current.gcvEarth.NAREC, digits=4))
Current.svmLinear.NAREC = mmetric(y=test.mean.Hours.24H$Current, x = test.mean.Hours.24H$pred.Current.svmLinear,
metric= "NAREC", aggregate="mean", val=5)
Current.svmLinear.NAREC.curve = paste("REC, svmLinear=", round(Current.svmLinear.NAREC, digits=4))
Current.brnn.NAREC = mmetric(y=test.mean.Hours.24H$Current, x = test.mean.Hours.24H$pred.Current.brnn,
metric= "NAREC", aggregate="mean", val=5)
Current.brnn.NAREC.curve = paste("REC, brnn=", round(Current.brnn.NAREC, digits=4))
Current.gam.NAREC = mmetric(y=test.mean.Hours.24H$Current, x = test.mean.Hours.24H$pred.Current.gam,
metric= "NAREC", aggregate="mean", val=5)
Current.gam.NAREC.curve = paste("REC, gam=", round(Current.gam.NAREC, digits=4))
#save to png #png looks nicer than the pdf
png("Current. REC-curves.png", width=1110, height=550)
split.screen(c(1,2))
screen(1)
L.graph<-mgraph(L, size=c(1,10), graph="REC",
leg=list(pos="bottomright",leg=c(Current.lm.NAREC.curve,
Current.cubist.NAREC.curve,
Current.bagEarthGCV.NAREC.curve,
Current.bagEarth.NAREC.curve)),
col=c("blue", "gray1", "green", "red"),
main="Current. REC curves")
screen(2)
B.graph<-mgraph(B, size=c(1,10), graph="REC",
leg=list(pos="bottomright",leg=c(Current.brnn.NAREC.curve,
Current.gcvEarth.NAREC.curve,
Current.svmLinear.NAREC.curve,
Current.gam.NAREC.curve)),
col=c("red", "gray1", "green", "blue"),
main="Current. REC curves")
close.screen(all=TRUE)
dev.off()
#Build the error metrics
errMetric.Current.brnn<-mmetric(y=test.mean.Hours.24H$Current, x = test.mean.Hours.24H$pred.Current.brnn,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.Current.bagEarth<-mmetric(y=test.mean.Hours.24H$Current, x = test.mean.Hours.24H$pred.Current.bagEarth,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.Current.bagEarthGCV<-mmetric(y=test.mean.Hours.24H$Current, x = test.mean.Hours.24H$pred.Current.bagEarthGCV,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.Current.cubist<-mmetric(y=test.mean.Hours.24H$Current, x = test.mean.Hours.24H$pred.Current.cubist,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.Current.lm<-mmetric(y=test.mean.Hours.24H$Current, x = test.mean.Hours.24H$pred.Current.lm,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.Current.gcvEarth<-mmetric(y=test.mean.Hours.24H$Current, x = test.mean.Hours.24H$pred.Current.gcvEarth,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.Current.svmLinear<-mmetric(y=test.mean.Hours.24H$Current, x = test.mean.Hours.24H$pred.Current.svmLinear,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
errMetric.Current.gam<-mmetric(y=test.mean.Hours.24H$Current, x = test.mean.Hours.24H$pred.Current.gam,
metric= c("RAE", "MSE", "ME","RMSE", "RSE", "RRSE", "R2", "Q2", "NAREC",
"MAPE", "MdAPE"), aggregate="mean")
#Combine all the metrics in one dataframe
errMatrix.Current<-data.frame(errMetric.Current.bagEarth,errMetric.Current.brnn,errMetric.Current.cubist,
errMetric.Current.lm,errMetric.Current.gcvEarth,errMetric.Current.svmLinear,
errMetric.Current.bagEarthGCV, errMetric.Current.gam)
errMatrix.Current<-t(errMatrix.Current)
#write.table(errMatrix.Current,"errorMatrixCurrent.csv", sep=",")
#Re-arrange the metric matrix with the most important accuracy indicators to the left