Ggtcf

Draw simple lines in Inkscape

What are these boxed doors outside store fronts in New York?

What does "enim et" mean?

How can the DM most effectively choose 1 out of an odd number of players to be targeted by an attack or effect?

I probably found a bug with the sudo apt install function

What is the offset in a seaplane's hull?

How to type dʒ symbol (IPA) on Mac?

"which" command doesn't work / path of Safari?

Why did the Germans forbid the possession of pet pigeons in Rostov-on-Don in 1941?

Why is "Reports" in sentence down without "The"

Can an x86 CPU running in real mode be considered to be basically an 8086 CPU?

My colleague's body is amazing

Should I join an office cleaning event for free?

How does one intimidate enemies without having the capacity for violence?

LED on same Pin as Toggle Switch, not illuminating

Is Social Media Science Fiction?

Download, install and reboot computer at night if needed

Is it possible to do 50 km distance without any previous training?

Why CLRS example on residual networks does not follows its formula?

Why do we use polarized capacitor?

Circuitry of TV splitters

Calculus Optimization - Point on graph closest to given point

Japan - Any leeway for max visa duration due to unforeseen circumstances?

What defenses are there against being summoned by the Gate spell?

in R, elastic net missing class in classification model

Remove rows with all or some NAs (missing values) in data.frameA simple explanation of Naive Bayes ClassificationErrors when running Caret package in RR and PCA Explanation for machine learningStatistical Learning--Classification model with LASSO prediction errorError using glmnet regression with metric as ROC in CaretError with class probabilities using R caretR - caret::train “random forest” parametersHow to adjust classification metrics to imbalanced binary classification data with `class_weight` in sklearnPredictive Analytics - “why” factor & model interpretability

.everyoneloves__top-leaderboard:empty,.everyoneloves__mid-leaderboard:empty,.everyoneloves__bot-mid-leaderboard:empty height:90px;width:728px;box-sizing:border-box;

0

In the cousera.org Machine Learning course assignment project, I adopt elastic net model to find the best classification model.

The training and testing data are compared to find what should be kept as training. I have checked NAs and highly correlated variables.

Data x is the training data, y is the variable "classe" in the train data. after removing highly correlated variables of x, 20 has been eliminated, which contains 32 numeric and no NAs variables.
Data y are factor "A"~"E".

elastic_model<-train(y=y,x=x,method="glmnet",trControl = trainControl("cv", number = 10),tuneLength=10)

above model got:

 elastic_model$bestTune

 alpha lambda

1 0.1 0.01217372

coef(elastic_model$finalModel,s=elastic_model$bestTune$lambda)

Above may give the results of all A~E classes, I can see some variables may be removed.

but below prediction give wrong results:

 predictions <- predict(elastic_model,testData)

 table(predictions,testData$classe)

predictions A B C D E

 A 108 47 19 16 37

 B 12 82 0 25 41

 C 1121 606 758 233 472

 D 433 404 249 690 532

 E 0 0 0 0 0

in the predictions, class E is lost totally. I can't find the reason.

r machine-learning classification r-caret

share|improve this question

edited Mar 9 at 4:55

NelsonGon

3,6594834

asked Mar 9 at 3:55

lovegirllovegirl

11

• 
  
  
  

  
  

  
  
  
  
  
  

  
  
  Please provide some sample data to work with. You can use dput for data.
  
  – NelsonGon
  Mar 9 at 4:57
  

  
  
  
  


• 
  
  
  

  
  

  
  
  
  
  
  

  
  
  Thank you! I have solved the problem by scaling the data.
  
  – lovegirl
  Mar 12 at 5:24
  

  
  
  
  

add a comment | 

0

In the cousera.org Machine Learning course assignment project, I adopt elastic net model to find the best classification model.

The training and testing data are compared to find what should be kept as training. I have checked NAs and highly correlated variables.

Data x is the training data, y is the variable "classe" in the train data. after removing highly correlated variables of x, 20 has been eliminated, which contains 32 numeric and no NAs variables.
Data y are factor "A"~"E".

elastic_model<-train(y=y,x=x,method="glmnet",trControl = trainControl("cv", number = 10),tuneLength=10)

above model got:

 elastic_model$bestTune

 alpha lambda

1 0.1 0.01217372

coef(elastic_model$finalModel,s=elastic_model$bestTune$lambda)

Above may give the results of all A~E classes, I can see some variables may be removed.

but below prediction give wrong results:

 predictions <- predict(elastic_model,testData)

 table(predictions,testData$classe)

predictions A B C D E

 A 108 47 19 16 37

 B 12 82 0 25 41

 C 1121 606 758 233 472

 D 433 404 249 690 532

 E 0 0 0 0 0

in the predictions, class E is lost totally. I can't find the reason.

r machine-learning classification r-caret

share|improve this question

edited Mar 9 at 4:55

NelsonGon

3,6594834

asked Mar 9 at 3:55

lovegirllovegirl

11

• 
  
  
  

  
  

  
  
  
  
  
  

  
  
  Please provide some sample data to work with. You can use dput for data.
  
  – NelsonGon
  Mar 9 at 4:57
  

  
  
  
  


• 
  
  
  

  
  

  
  
  
  
  
  

  
  
  Thank you! I have solved the problem by scaling the data.
  
  – lovegirl
  Mar 12 at 5:24
  

  
  
  
  

add a comment | 

In the cousera.org Machine Learning course assignment project, I adopt elastic net model to find the best classification model.

The training and testing data are compared to find what should be kept as training. I have checked NAs and highly correlated variables.

Data x is the training data, y is the variable "classe" in the train data. after removing highly correlated variables of x, 20 has been eliminated, which contains 32 numeric and no NAs variables.
Data y are factor "A"~"E".

elastic_model<-train(y=y,x=x,method="glmnet",trControl = trainControl("cv", number = 10),tuneLength=10)

above model got:

 elastic_model$bestTune

 alpha lambda

1 0.1 0.01217372

coef(elastic_model$finalModel,s=elastic_model$bestTune$lambda)

Above may give the results of all A~E classes, I can see some variables may be removed.

but below prediction give wrong results:

 predictions <- predict(elastic_model,testData)

 table(predictions,testData$classe)

predictions A B C D E

 A 108 47 19 16 37

 B 12 82 0 25 41

 C 1121 606 758 233 472

 D 433 404 249 690 532

 E 0 0 0 0 0

in the predictions, class E is lost totally. I can't find the reason.

r machine-learning classification r-caret

share|improve this question

edited Mar 9 at 4:55

NelsonGon

3,6594834

asked Mar 9 at 3:55

lovegirllovegirl

11

elastic_model<-train(y=y,x=x,method="glmnet",trControl = trainControl("cv", number = 10),tuneLength=10)

 elastic_model$bestTune

 alpha lambda

1 0.1 0.01217372

coef(elastic_model$finalModel,s=elastic_model$bestTune$lambda)

 predictions <- predict(elastic_model,testData)

 table(predictions,testData$classe)

predictions A B C D E

 A 108 47 19 16 37

 B 12 82 0 25 41

 C 1121 606 758 233 472

 D 433 404 249 690 532

 E 0 0 0 0 0

share|improve this question

edited Mar 9 at 4:55

NelsonGon

3,6594834

asked Mar 9 at 3:55

lovegirllovegirl

11

share|improve this question

edited Mar 9 at 4:55

NelsonGon

3,6594834

asked Mar 9 at 3:55

lovegirllovegirl

11

edited Mar 9 at 4:55

NelsonGon

3,6594834

edited Mar 9 at 4:55

NelsonGon

3,6594834

asked Mar 9 at 3:55

lovegirllovegirl

11

asked Mar 9 at 3:55

lovegirllovegirl

11