I would like to use glmnet package for classification project. I managed to build the model and get the prdiction per row but I don't see a way to use it as a regular variable. How can I add the prediction vector to the data frame as a regular column that I'll be able to work upon (like calculate its mean etc..) Here is my toy data frame and code.Also, I see that glmnet called the prediction vector as "1", why is that nad how can I change it?
df <- read.table(text = "target birds wolfs
1 9 7
1 8 4
0 2 8
1 2 3
1 8 3
0 1 2
1 7 1
0 1 5
1 9 7
1 8 7
0 2 7
0 2 3
1 6 3
0 1 1
0 3 9
0 1 1 ",header = TRUE)
library(Matrix)
some_matrix <- data.matrix(df[,2:3])
Matrix(some_matrix, sparse=TRUE)
set.seed(2)
split <- df[1:10,]
split <- NULL
train <- df[1:10,]
test<-df[11:16,]
train_sparse <- sparse.model.matrix(~.,train[2:3])
test_sparse <- sparse.model.matrix(~.,test[2:3])
library(glmnet)
fit <- glmnet(train_sparse,train[,1],family='binomial')
cv <- cv.glmnet(train_sparse,train[,1],nfolds=3)
pred <- predict(fit, test_sparse,type="response", s=cv$lambda.min)
pred
1
11 0.09691107
12 0.82760730
13 1.00000000
14 0.26871408
15 0.58367863
16 0.26871408
c<-cbind(test,pred)
target birds wolfs 1
11 0 2 7 0.09691107
12 0 2 3 0.82760730
13 1 6 3 1.00000000
14 0 1 1 0.26871408
15 0 3 9 0.58367863
16 0 1 1 0.26871408
typeof(c$1)
Error: unexpected numeric constant in "typeof(c$1"
mean(c$1)
Error: unexpected numeric constant in "mean(c$1"
> str(c)
'data.frame': 6 obs. of 5 variables:
$ target: int 0 0 1 0 0 0
$ birds : int 2 2 6 1 3 1
$ wolfs : int 7 3 3 1 9 1
$ 1 : num 0.0969 0.8276 1 0.2687 0.5837 ...
$ pred : num [1:6, 1] 0.0969 0.8276 1 0.2687 0.5837 ...
..- attr(*, "dimnames")=List of 2
.. ..$ : chr "11" "12" "13" "14" ...
.. ..$ : chr "1"
df <- read.table(text = "target birds wolfs
1 9 7
1 8 4
0 2 8
1 2 3
1 8 3
0 1 2
1 7 1
0 1 5
1 9 7
1 8 7
0 2 7
0 2 3
1 6 3
0 1 1
0 3 9
0 1 1 ",header = TRUE)
set.seed(44)
library(Matrix)
library(glmnet)
some_matrix <- data.matrix(df[,2:3])
Matrix(some_matrix, sparse=TRUE)
set.seed(2)
split <- df[1:10,]
split <- NULL
train <- df[1:10,]
test<-df[11:16,]
train_sparse <- model.matrix(~.,train[2:3]) #changed to model.matrix
test_sparse <- model.matrix(~.,test[2:3])
library(glmnet)
fit <- glmnet(train_sparse,train[,1],family='binomial')
cv <- cv.glmnet(train_sparse,train[,1],nfolds=3) # i believe you can do all of the modeling in here. You should skip the "fit" part.
pred <- predict(fit, test_sparse,type="response", s=cv$lambda.min)
test_sparse<-as.data.frame(test_sparse[,-1]) #removing intercept
test_sparse$predictionColumn<-c(pred) #adding in new column