classification/ILPD_OverSampling.R at main · mehcavdar/classification · GitHub

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
--------------------------------------------------------------------------------------------------
  load(file = "LPIDDataFrame.rda")
  load(file="LPIDPerformance.rda")
--------------------------------------------------------------------------------------------------
  # kütüphaneleri yükle ####

library(dplyr)
library(caret)
library(GGally)
library(plyr)
library(ROSE)
library(class)
library(magrittr)
library(DMwR)
  #####################################################
# Egitim Veri Seti üzerinde  (Oversampling Yontemi) -Minority instance increase
#####################################################


#gozlemSayisi <-(as.numeric(table(egitimVS$is_patient )[2]) * 2 )
gozlemSayisi <-(as.numeric(table(df$is_patient )[2]) * 2 )
#overVeriSeti <-ovun.sample(formula=is_patient~., data=egitimVS, N=gozlemSayisi , seed=1 , method="over")$data
overVeriSeti <-ovun.sample(formula=is_patient~., data=df, N=gozlemSayisi , seed=1 , method="over")$data
overVeriSeti$is_patient  <- relevel(overVeriSeti$is_patient,ref = "NO")
table(overVeriSeti$is_patient)

###############################################
# egitim ve test veri setinin Olusturlmasi
##############################################


set.seed(1)
egitimIndisleri <- createDataPartition(y = overVeriSeti$is_patient, p = .70, list = FALSE)
egitimVS <- overVeriSeti[egitimIndisleri,]
testVS   <- overVeriSeti[-egitimIndisleri,]

table(egitimVS$is_patient)
table(testVS$is_patient)

#########################################################
# MOdelleme# k-En Yakin Komsu Algoritmasinin uygulanmasi
##########################################################
set.seed(1)
(tahminler_over_veriSeti <- knn(train = overVeriSeti[, -9], test = testVS[, -9], cl = overVeriSeti[[9]], k = 5))
tahminler_over_veriSeti
testVS[[9]]

###############################################
# Performans degerlendirmesi
###############################################

(tablom <- table(tahminler_over_veriSeti, testVS[[9]], dnn = c("Tahmini Siniflar", "Gercek Siniflar")))
cm <- confusionMatrix(data = tahminler_over_veriSeti, reference = testVS[[9]], mode = "everything" )
cm
degerlerOver<-c("OverVS",cm$overall[1],cm$byClass[1],cm$byClass[2],cm$byClass[3],cm$byClass[7])
names(degerlerOver)<-c("OverVS","Dogruluk","Duyarlilik","Belirleyicilik","Kesinlik","F-Ölçütü")
degerlerEgitim
degerlerOver

save(degerlerEgitim,degerlerOver, file="LPIDPerformance.rda")