Skip to main content
Cross Validated

Return to Question

added 33 characters in body
Source Link
uared1776
uared1776
  • 161
  • 1
  • 13

Can feature Feature selection using prior knowledge introduce bias?

Suppose I have a set of predictors, for a regression problem. I know some of them maybe useless, but I am not sure.

So, I build multiple versions of predictor set, that eacheach version containcontains/not containcontains some of the predictors that I am not sure about. Then, for each version, I use cross-validation (CV) to tune a same learning model. I then calculate the CV error.

Can I use the predictor set with smallest CV error? Does it introduce bias or overfitting? Why?

Will it help, if a separate test set is used?

Can feature selection using prior knowledge introduce bias?

Suppose I have a set of predictors, I know some of them maybe useless, but I am not sure.

So, I build multiple versions of predictor set, that each version contain/not contain some of the predictors that I am not sure about. Then, for each version, I use cross-validation (CV) to tune a same learning model. I then calculate the CV error.

Can I use the predictor set with smallest CV error? Does it introduce bias or overfitting? Why?

Will it help, if a separate test set is used?

Feature selection using prior knowledge?

Suppose I have a set of predictors for a regression problem. I know some of them maybe useless, but I am not sure.

So, I build multiple versions of predictor set, that each version contains/not contains some of the predictors that I am not sure about. Then, for each version, I use cross-validation (CV) to tune a same learning model. I then calculate the CV error.

Can I use the predictor set with smallest CV error? Does it introduce bias or overfitting? Why?

Will it help, if a separate test set is used?

added 47 characters in body
Source Link
uared1776
uared1776
  • 161
  • 1
  • 13

Suppose I have a set of predictors, I know some of them maybe useless, but I am not sure.

So, I build multiple versions of predictor set, that each version contain/not contain some of the predictors that I am not sure about. Then, for each version, I use cross-validation (CV) to tune a same learning model. I then calculate the CV error.

Can I use the predictor set with smallest CV error? Does it introduce bias or overfitting? Why?

Will it help, if a separate test set is used?

Suppose I have a set of predictors, I know some of them maybe useless, but I am not sure.

So, I build multiple versions of predictor set, that each version contain/not contain some of the predictors that I am not sure about. Then, for each version, I use cross-validation (CV) to tune a same learning model. I then calculate the CV error.

Can I use the predictor set with smallest CV error? Does it introduce bias or overfitting? Why?

Suppose I have a set of predictors, I know some of them maybe useless, but I am not sure.

So, I build multiple versions of predictor set, that each version contain/not contain some of the predictors that I am not sure about. Then, for each version, I use cross-validation (CV) to tune a same learning model. I then calculate the CV error.

Can I use the predictor set with smallest CV error? Does it introduce bias or overfitting? Why?

Will it help, if a separate test set is used?

Source Link
uared1776
uared1776
  • 161
  • 1
  • 13

Can feature selection using prior knowledge introduce bias?

Suppose I have a set of predictors, I know some of them maybe useless, but I am not sure.

So, I build multiple versions of predictor set, that each version contain/not contain some of the predictors that I am not sure about. Then, for each version, I use cross-validation (CV) to tune a same learning model. I then calculate the CV error.

Can I use the predictor set with smallest CV error? Does it introduce bias or overfitting? Why?