CODEWITHSUNDEEP
rdataframevector
Yehuda
Yehuda

Reputation: 1893

"argument is not numeric or logical: returning NA" error when using function mean [R]

I used the Ames data set to create the following code:

NbrMLR <- lm(SalePrice ~ Neighborhood, data = ames_housing_data)
pred <- as.data.frame(predict(NbrMLR, ames_housing_data))
library(reshape)
pred <- rename(pred,c('predict(NbrMLR, ames_housing_data)' = 'prd'))
ames_housing_data$NbrPred <- pred
ames_housing_data$NbrRes <- SalePrice - ames_housing_data$NbrPred
ames_housing_data$absoluteNbrRes <- abs(ames_housing_data$NbrRes)
NbrMAE <- mean(ames_housing_data$absoluteNbrRes)

This code results in the following error:

argument is not numeric or logical: returning NA

str(ames_housing_data$absoluteNbrRes) tells that I have a data frame of variable type $ prd: num. Though I'm familiar with num, I've never heard of prd: num before. Does this cause problem? Why does not mean() act upon what otherwise seems to me to be a valid vector?

Upvotes: 2

Views: 187

Answers (1)

Zheyuan Li
Zheyuan Li

Reputation: 73265

ames_housing_data$NbrPred <- pred[[1]] will fix it.

Regarding your error: prd is a variable name (as you defined it!). You pass a data frame to mean, which causes you error.

If I were you, I would write the following code:

NbrMLR <- lm(SalePrice ~ Neighborhood, data = ames_housing_data)
ames_housing_data$NbrPred <- predict(NbrMLR, ames_housing_data)
ames_housing_data$NbrRes <- SalePrice - ames_housing_data$NbrPred
ames_housing_data$absoluteNbrRes <- abs(ames_housing_data$NbrRes)
NbrMAE <- mean(ames_housing_data$absoluteNbrRes)

Upvotes: 1

Related Questions