-
Notifications
You must be signed in to change notification settings - Fork 0
/
FinalProject.Rmd
1170 lines (874 loc) · 44.2 KB
/
FinalProject.Rmd
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
---
title: "Final Project"
author: "Lyndon Swarey, Ryan Folk, Alexa Kelly"
date: "May 7, 2020"
output:
pdf_document: default
html_document: default
---
```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```
## Packages
The following packages are required. Tidyverse is used for data manipulation, cluster is used for k-means clustering, factoextra is used for visualizing the cluster results, caret package is used to split the data, and fit the K-NN model, rpart is used to fit the classification tree and visualize the results, MASS is used to fit the linear and quadratic discriminant analysis, reshape2 is used to convert the data from wide to long and "select <- dplyr::select" avoids conflict with the MASS package.
```{r, warning=FALSE, message=FALSE}
library(tidyverse)
library(cluster)
library(factoextra)
library(caret)
library(rpart)
library(MASS)
library(reshape2)
select<-dplyr::select
```
## Data
The data was loaded using read.csv. To preserve the originial dataset, a new dataset was created from the original. A new ID variable was also created.
```{r}
# Import data
data_org <- read.csv("/Users/alexa/Desktop/data.csv")
# Create a new file to preserve orgininal data
data <- data_org
# Create New_Id column
data <- data %>%
mutate(new_ID = seq(1:569))
```
## Check for Missing Values
The following code checked for missing values. This data set does not have any missing values
```{r}
anyNA(data)
```
## Variable clustering: mean
### radius_mean
The radius_mean variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=3 with 10 random starts. A new variable in data_3 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select radius_mean
data_3 <- data %>%
select(radius_mean)
# Set seed for reproducibility
set.seed(123)
# Graph for wws method
fviz_nbclust(data_3, kmeans, method = "wss") +
labs(subtitle = "Elbow method: radius_mean")
# Kmeans wss method
k_wss <- kmeans(data_3, centers = 3, nstart = 10)
# Make new colums in for the new variables generated from kmeans
data_3 <- data_3 %>%
mutate(radius_mean_wss = as.factor(as.factor(k_wss$cluster)),
new_ID = seq(1:569))
```
### texture_mean
The texture_mean variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=3 with 10 random starts. A new variable in data_4 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select texture_mean
data_4 <- data %>%
select(texture_mean)
# Set seed for reproducibility
set.seed(123)
# Graph for wws method
fviz_nbclust(data_4, kmeans, method = "wss") +
labs(subtitle = "Elbow method: texture_mean")
# Kmeans wss method
k_wss <- kmeans(data_4, centers = 3, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_4 <- data_4 %>%
mutate(texture_mean_wss = as.factor(as.factor(k_wss$cluster)),
new_ID = seq(1:569))
```
### perimeter_mean
The perimeter_mean variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=3 with 10 random starts. A new variable in data_5 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select perimeter_mean
data_5 <- data %>%
select(perimeter_mean)
# Set seed for reproducibilty
set.seed(123)
# Graph for wws method
fviz_nbclust(data_5, kmeans, method = "wss") +
labs(subtitle = "Elbow method: perimeter_mean")
# Kmeans wss method
k_wss <- kmeans(data_5, centers = 3, nstart = 10)
# Make new colums in data_3 for the new variables generated from kmeans
data_5 <- data_5 %>%
mutate(perimeter_mean_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### area_mean
The area_mean variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=3 with 10 random starts. A new variable in data_6 was created for the cluster results and a new_ID variable was made as well. The optimal k appears to be at 4 but needed to be adjusted in order maintain the minimum number of observations for each catorgory for the Chi-square test.
```{r}
# Select area_mean
data_6 <- data %>%
select(area_mean)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_6, kmeans, method = "wss") +
labs(subtitle = "Elbow method: area_mean")
# Kmeans wss method
k_wss <- kmeans(data_6, centers = 3, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_6 <- data_6 %>%
mutate(area_mean_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### smoothness_mean
The smoothness_mean variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=3 with 10 random starts. A new variable in data_7 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select smoothness_mean
data_7 <- data %>%
select(smoothness_mean)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_7, kmeans, method = "wss") +
labs(subtitle = "Elbow method: smoothness_mean")
# Kmeans wss method
k_wss <- kmeans(data_7, centers = 3, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_7 <- data_7 %>%
mutate(smoothness_mean_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### compactness_mean
The compactness_mean variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=3 with 10 random starts. A new variable in data_8 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select smoothness_mean
data_8 <- data %>%
select(compactness_mean)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_8, kmeans, method = "wss") +
labs(subtitle = "Elbow method: compactness_mean")
# Kmeans wss method
k_wss <- kmeans(data_8, centers = 3, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_8 <- data_8 %>%
mutate(compactness_mean_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### concavity_mean
The concavity_mean variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=3 with 10 random starts. A new variable in data_9 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select concavity_mean
data_9 <- data %>%
select(concavity_mean)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_9, kmeans, method = "wss") +
labs(subtitle = "Elbow method: concavity_mean")
# Kmeans wss method
k_wss <- kmeans(data_9, centers = 3, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_9 <- data_9 %>%
mutate(concavity_mean_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### concave.points_mean
The concave.points_mean variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=3 with 10 random starts. A new variable in data_10 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select concave.points_mean
data_10 <- data %>%
select(concave.points_mean)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_10, kmeans, method = "wss") +
labs(subtitle = "Elbow method: concave.points_mean")
# Kmeans wss method
k_wss <- kmeans(data_10, centers = 3, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_10 <- data_10 %>%
mutate(concave.points_mean_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### symmetry_mean
The symmetry_mean variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=3 with 10 random starts. A new variable in data_11 was created for the cluster results and a new_ID variable was made as well. The optimal k appears to be at 4 but needed to be adjusted in order maintain the minimum number of observations for each catorgory for the Chi-square test.
```{r}
# Select symmetry_mean
data_11 <- data %>%
select(symmetry_mean)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_11, kmeans, method = "wss") +
labs(subtitle = "Elbow method: symmetry_mean")
# Kmeans wss method
k_wss <- kmeans(data_11, centers = 3, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_11 <- data_11 %>%
mutate(symmetry_mean_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### fractal_dimension_mean
The fractal_dimension_mean variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=4 with 10 random starts. A new variable in data_12 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select fractal_dimension_mean
data_12 <- data %>%
select(fractal_dimension_mean)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_12, kmeans, method = "wss") +
labs(subtitle = "Elbow method: fractal_dimension_mean")
# Kmeans wss method
k_wss <- kmeans(data_12, centers = 3, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_12 <- data_12 %>%
mutate(fractal_dimension_mean_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
## Variable clustering: se
### radius_se
The radius_se variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=3 with 10 random starts. A new variable in data_13 was created for the cluster results and a new_ID variable was made as well. The optimal k appears to be at 4 but needed to be adjusted in order maintain the minimum number of observations for each catorgory for the Chi-square test.
```{r}
# Select radius_se
data_13 <- data %>%
select(radius_se)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_13, kmeans, method = "wss") +
labs(subtitle = "Elbow method: radius_se")
# Kmeans wss method
k_wss <- kmeans(data_13, centers = 3, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_13 <- data_13 %>%
mutate(radius_se_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### texture_se
The texture_se variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=4 with 10 random starts. A new variable in data_14 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select texture_se
data_14 <- data %>%
select(texture_se)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_14, kmeans, method = "wss") +
labs(subtitle = "Elbow method: texture_se")
# Kmeans wss method
k_wss <- kmeans(data_14, centers = 3, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_14 <- data_14 %>%
mutate(texture_se_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### perimeter_se
The perimeter_se variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=3 with 10 random starts. A new variable in data_15 was created for the cluster results and a new_ID variable was made as well. The optimal k appears to be at 4 but needed to be adjusted in order maintain the minimum number of observations for each catorgory for the Chi-square test.
```{r}
# Select perimeter_se
data_15 <- data %>%
select(perimeter_se)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_15, kmeans, method = "wss") +
labs(subtitle = "Elbow method: perimeter_se")
# Kmeans wss method
k_wss <- kmeans(data_15, centers = 3, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_15 <- data_15 %>%
mutate(perimeter_se_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### area_se
The area_se variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=2 with 10 random starts. A new variable in data_16 was created for the cluster results and a new_ID variable was made as well. The optimal k appears to be at 3 but needed to be adjusted in order maintain the minimum number of observations for each catorgory for the Chi-square test.
```{r}
# Select area_se
data_16 <- data %>%
select(area_se)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_16, kmeans, method = "wss") +
labs(subtitle = "Elbow method: area_se")
# Kmeans wss method
k_wss <- kmeans(data_16, centers = 2, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_16 <- data_16 %>%
mutate(area_se_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### smoothness_se
The smoothness_se variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=3 with 10 random starts. A new variable in data_17 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select smoothness_se
data_17 <- data %>%
select(smoothness_se)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_17, kmeans, method = "wss") +
labs(subtitle = "Elbow method: smoothness_se")
# Kmeans wss method
k_wss <- kmeans(data_17, centers = 3, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_17 <- data_17 %>%
mutate(smoothness_se_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### compactness_se
The compactnes_se variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=3 with 10 random starts. A new variable in data_18 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select compactness_se
data_18 <- data %>%
select(compactness_se)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_18, kmeans, method = "wss") +
labs(subtitle = "Elbow method: compactness_se")
# Kmeans wss method
k_wss <- kmeans(data_18, centers = 3, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_18 <- data_18 %>%
mutate(compactness_se_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### concavity_se
The concavity_se variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=2 with 10 random starts. A new variable in data_19 was created for the cluster results and a new_ID variable was made as well. The optimal k appears to be at 4 but needed to be adjusted in order maintain the minimum number of observations for each catorgory for the Chi-square test.
```{r}
# Select concavity_se
data_19 <- data %>%
select(concavity_se)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_19, kmeans, method = "wss") +
labs(subtitle = "Elbow method: concavity_se")
# Kmeans wss method
k_wss <- kmeans(data_19, centers = 2, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_19 <- data_19 %>%
mutate(concavity_se_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### concave.points_se
The concave.points_se variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=4 with 10 random starts. A new variable in data_20 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select concave.points_se
data_20 <- data %>%
select(concave.points_se)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_20, kmeans, method = "wss") +
labs(subtitle = "Elbow method: concave.points_se")
# Kmeans wss method
k_wss <- kmeans(data_20, centers = 4, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_20 <- data_20 %>%
mutate(concave.points_se_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### symmetry_se
The symmetry_se variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=4 with 10 random starts. A new variable in data_21 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select symetry_se
data_21 <- data %>%
select(symmetry_se)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_21, kmeans, method = "wss") +
labs(subtitle = "Elbow method: symmetry_se")
# Kmeans wss method
k_wss <- kmeans(data_21, centers = 4, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_21 <- data_21 %>%
mutate(symmetry_se_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### fractal_dimension_se
The fractal_dimension_se variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=3 with 10 random starts. A new variable in data_22 was created for the cluster results and a new_ID variable was made as well. The optimal k appears to be at 4 but needed to be adjusted in order maintain the minimum number of observations for each catorgory for the Chi-square test.
```{r}
# Select fractal_dimension_mean
data_22 <- data %>%
select(fractal_dimension_se)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_22, kmeans, method = "wss") +
labs(subtitle = "Elbow method: fractal_dimension_se")
# Kmeans wss method
k_wss <- kmeans(data_22, centers = 3, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_22 <- data_22 %>%
mutate(fractal_dimension_se_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
## Variable clustering: worst
### radius_worst
The radius_worst variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=3 with 10 random starts. A new variable in data_23 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select radius_worst
data_23 <- data %>%
select(radius_worst)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_23, kmeans, method = "wss") +
labs(subtitle = "Elbow method: radius_worst")
# Kmeans wss method
k_wss <- kmeans(data_23, centers = 3, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_23 <- data_23 %>%
mutate(radius_worst_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### texture_worst
The texture_worst variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=4 with 10 random starts. A new variable in data_24 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select texture_worst
data_24 <- data %>%
select(texture_worst)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_24, kmeans, method = "wss") +
labs(subtitle = "Elbow method: texture_worst")
# Kmeans wss method
k_wss<- kmeans(data_24, centers = 3, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_24 <- data_24 %>%
mutate(texture_worst_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### perimeter_worst
The perimeter_worst variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=4 with 10 random starts. A new variable in data_25 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select perimeter_worst
data_25 <- data %>%
select(perimeter_worst)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_25, kmeans, method = "wss") +
labs(subtitle = "Elbow method: perimeter_worst")
# Kmeans wss method
k_wss <- kmeans(data_25, centers = 3, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_25 <- data_25 %>%
mutate(perimeter_worst_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### area_worst
The area_worst variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=4 with 10 random starts. A new variable in data_26 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select area_worst
data_26 <- data %>%
select(area_worst)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_26, kmeans, method = "wss") +
labs(subtitle = "Elbow method: area_worst")
# Kmeans wss method
k_wss <- kmeans(data_26, centers = 3, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_26 <- data_26 %>%
mutate(area_worst_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### smoothness_worst
The smoothness_worst variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=4 with 10 random starts. A new variable in data_27 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select smoothness_worst
data_27 <- data %>%
select(smoothness_worst)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_27, kmeans, method = "wss") +
labs(subtitle = "Elbow method: smoothness_worst")
# Kmeans wss method
k_wss <- kmeans(data_27, centers = 4, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_27 <- data_27 %>%
mutate(smoothness_worst_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### compactness_worst
The compactness_worst variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=4 with 10 random starts. A new variable in data_28 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select compactness_worst
data_28 <- data %>%
select(compactness_worst)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_28, kmeans, method = "wss") +
labs(subtitle = "Elbow method: compactness_worst")
# Kmeans wss method
k_wss <- kmeans(data_28, centers = 4, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_28 <- data_28 %>%
mutate(compactness_worst_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### concavity_worst
The concavity_worst variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=3 with 10 random starts. A new variable in data_29 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select radius_worst
data_29 <- data %>%
select(concavity_worst)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_29, kmeans, method = "wss") +
labs(subtitle = "Elbow method: concavity_worst")
# Kmeans wss method
k_wss <- kmeans(data_29, centers = 3, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_29 <- data_29 %>%
mutate(concavity_worst_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### concave.points_worst
The concave.points_worst variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=4 with 10 random starts. A new variable in data_30 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select concave.points_worst
data_30 <- data %>%
select(concave.points_worst)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_30, kmeans, method = "wss") +
labs(subtitle = "Elbow method: concave.points_worst")
# Kmeans wss method
k_wss <- kmeans(data_30, centers = 4, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_30 <- data_30 %>%
mutate(concave.points_worst_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### symmetry_worst
The symmetry_worst variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=4 with 10 random starts. A new variable in data_31 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select radius_worst
data_31 <- data %>%
select(symmetry_worst)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_31, kmeans, method = "wss") +
labs(subtitle = "Elbow method: symmetry_worst")
# Kmeans wss method
k_wss <- kmeans(data_31, centers = 3, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_31 <- data_31 %>%
mutate(symmetry_worst_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
### fractal_dimension_worst
The fractal_dimension_worst variable was selected and the seed set for reproduciblity. A graph was created for the different clusters using the within sum of squares for k=1 to k=10. The optimal k was found using the elbow method. The k-means algorithm was run using the optimal k=4 with 10 random starts. A new variable in data_32 was created for the cluster results and a new_ID variable was made as well.
```{r}
# Select radius_worst
data_32 <- data %>%
select(fractal_dimension_worst)
# Set seed
set.seed(123)
# Graph for wws method
fviz_nbclust(data_32, kmeans, method = "wss") +
labs(subtitle = "Elbow method: fractal_dimension_worst")
# Kmeans wss method
k_wss <- kmeans(data_32, centers = 4, nstart = 10)
# Make new colums for the new variables generated from kmeans
data_32 <- data_32 %>%
mutate(fractal_dimension_worst_wss = as.factor(k_wss$cluster),
new_ID = seq(1:569))
```
## Merge Datasets
Id, diagnosis and new_ID were selected and a merge function was made that allowed multiple data frames to be merged at the same time. The the different data frames were merged by the new function and the columns in data_new were rearranged to have all the origninal variables together and all the variables ending in wss together.
```{r}
#Select the first 2 row the data file
data_new <- data %>%
select(c(1,2,33))
# Make a merge function
mymerge <- function(x, y){
df <- merge(x, y, by= "new_ID", all.x= TRUE, all.y= TRUE)
return(df)}
# Merge files
data_new <-Reduce( mymerge, list(data_new, data_3,data_4, data_5, data_6,
data_7, data_8, data_9, data_10, data_11,
data_12, data_13, data_14, data_15, data_16,
data_17, data_18, data_19, data_20, data_21,
data_22, data_23, data_24, data_25, data_26,
data_27, data_28, data_29, data_30, data_31, data_32))
# Arrange the columns
data_new <- data_new[c(1,2,3,4,6,8,10,12,14,16,18,20,
22,24,26,28,30,32,34,36,38,40,
42,44,46,48,50,52,54,56,58,60,62,
5,7,9,11,13,15,17,19,21,23,25,27,
29,31,33,35,37,39,41,43,45,47,
49,51,53,55,57,59,61,63)]
```
## Scale Continous Variables
The continues variables were selected and scaled to have a mean of zero and a standard deviation of one. These variables were then removed leaving only the new scaled variables and the new_ID variable was created. data_new_scale was then merged with data_new by new_ID.
```{r}
# Rescale Variables to have a mean of 0 and standard deviation of 1
data_new_scale <-data_new %>%
select(c(4:33)) %>%
mutate_all(funs(scale = ((.)-mean(.))/sd(.))) %>%
select(-c(1:30)) %>%
mutate(new_ID = seq(c(1:569)))
# Merge data_new_scale with data_new
data_new <- merge(data_new, data_new_scale, by = "new_ID")
```
## Transform Variables to Make Them Normal
The continuous variables were selected and converted to long form. A histogram was made of the variables to determine if they were were normally distributed. Some of the variables appear to not be normally distributed. The continuous variables were reselected, the variables to transformed were created and the original variables were removed. A Box Cox transformation was applied to these variables. The data was converted to long form again and a histogram of the variables was made. A few of the variables have outliers. Upon farther investigation, these outliers do not appear to be typos, but legitimate data and will be kept as part of the analysis. This may affect the results of the lda and qda models. new_ID variable was created and data_new_trans was merged with data_new by the new_ID variable.
```{r, warning=FALSE, message=FALSE}
# Select the continous variables
data_new_trans <-data_new %>%
select(c(4:33))
# Convert to long format
melt_df<-melt(data_new_trans)
# Histogram of all variables
ggplot(melt_df, aes(x=value)) +
geom_histogram()+
facet_wrap(~variable, scale="free")+
labs(title = "Histogram of Continuous Variables")
# Select continous variables to be transformed
data_new_trans <-data_new %>%
select(c(4:33)) %>%
mutate_all(funs(trans = (.))) %>%
select(-c(1:30))
# Apply Box Cox transformation
process_vars <- preProcess(data_new_trans, method = c("BoxCox"))
data_new_trans <- predict(process_vars, data_new_trans)
# Convert to long format
melt_df<-melt(data_new_trans)
# Histogram of transformed variables
ggplot(melt_df, aes(x=value)) +
geom_histogram()+
facet_wrap(~variable, scale="free")+
labs(title = "Histogram of Transformed Variables")
# Create new variable to merge on
data_new_trans <- data_new_trans %>%
mutate(new_ID = seq(c(1:569)))
# Merge data_new_trans with data_new
data_new <- merge(data_new, data_new_trans, by = "new_ID")
```
## Variable Selection and Data Processing
### Split Dataset With All Variables
A seed was set for reproducibility and the data was split 70:30 into training and test datasets.
```{r}
# Set seed for reproducibility
set.seed(123)
# Split data into training and test datasets
index <- createDataPartition(data_new$diagnosis,
p = 0.7,
list = FALSE)
train_data <- data_new[index, ]
test_data <- data_new[-index, ]
```
### Chi-square Test
The binned variables along with diagnosis were selected. A special function was made to run the Chi-square test on the all the variables at once. The statistically significant variable were selected. A seed was set for reproducibility and the dataset including only the variables selected by the chi-square test was split 70:30 into training and test datasets.
```{r}
# Select the variable to be use in the Chi-square test
data_chisquare <- data_new %>%
select(c(34:63,3))
# Run the Chi-square test
myTests <- lapply(data_chisquare[-length(data_chisquare)], function(x) chisq.test(table(as.factor(x), data_chisquare$diagnosis)))
unlist(sapply(myTests, "[", "p.value"))
# Select the statistically significant variables at .001
data_chisquare <- data_new %>%
select(-c(smoothness_se_wss, concavity_se_wss,
symmetry_se_wss, fractal_dimension_mean_wss,
texture_se_wss, fractal_dimension_se_wss,
smoothness_se, concavity_se, symmetry_se,
fractal_dimension_mean, texture_se, fractal_dimension_se,
smoothness_se_scale, concavity_se_scale,
symmetry_se_scale, fractal_dimension_mean_scale,
texture_se_scale, fractal_dimension_se_scale,
smoothness_se_trans, concavity_se_trans,
symmetry_se_trans, fractal_dimension_mean_trans,
texture_se_trans, fractal_dimension_se_trans))
# Set seed for reproducibility
set.seed(123)
# Split data into training and test datasets
index <- createDataPartition(data_chisquare$diagnosis,
p = 0.7,
list = FALSE)
train_chisquare <- data_chisquare[index, ]
test_chisquare <- data_chisquare[-index, ]
```
### Classification Tree
A seed was set for reproducibility, and a classification tree was run using the default settings of the rpart() function. The resulting tree was viewed and the variables that were selected by the tree were placed into a new dataset. A seed was set again for reproducibility and this dataset was partition 70:30 into training and test datasets.
```{r}
# Set seed for reproducibility
set.seed(123)
# Fit the classification tree model
data_classtree <- rpart(diagnosis ~ ., data = data_new,
method = "class")
#plot tree
library(RColorBrewer)
library(rattle)
fancyRpartPlot(data_classtree, palette="RdYlGn", sub=" ")
```
```{r }
# View the variables selected
print(data_classtree)
# Select the variables
data_classtree <- data_new %>%
select(c(diagnosis, radius_worst, radius_worst_wss, radius_worst_scale, radius_worst_trans,
concave.points_worst, concave.points_worst_wss, concave.points_worst_scale,
concave.points_worst_trans, texture_worst, texture_worst_wss,
texture_worst_scale, texture_worst_trans))
# Set seed for reproducibility
set.seed(123)
# Split data into training and test datasets
index <- createDataPartition(data_classtree$diagnosis,
p = 0.7,
list = FALSE)
train_classtree <- data_classtree[index, ]
test_classtree <- data_classtree[-index, ]
```
## K-NN
### All Variables
K-NN was run on all the variables. The scaled variables were selected and a seed set for reproducibility. The model was fit to the training dataset with 10-fold cross-validation. The resulting model was used to classify the type of cancer on the test dataset and a confusion matrix was generated.
```{r}
# Select scaled variables
train_data_knn <- train_data %>%
select(diagnosis, ends_with("scale"))
# Set seed for reproducibility
set.seed(123)
# Fit the model
knn_fit = train(diagnosis~.,
data = train_data_knn,
method = "knn",
trControl = trainControl(method = "cv", number = 10))
# Make predictions using the test dataset
knnPredict = predict(knn_fit, newdata = test_data)
# Confusion matrix
confusionMatrix(knnPredict, test_data$diagnosis, positive = "M")
```
### Chi-square Variables
K-NN was run on the variables selected by the Chi-square test. The scaled variables were selected and a seed set for reproducibility. The model was fit to the training dataset with 10-fold cross-validation. The resulting model was used to classify the type of cancer on the test dataset, and a confusion matrix was generated.
```{r}