mean returns NaN besides na.rm= TRUE

Question

Sample data

date        coins   
2013-10-01  NA      
2013-10-01  NA      
2013-10-01  NA      
2013-11-01  10      
2013-11-01  NA      
2013-11-01  20      
2013-11-01  30      
2013-11-01  40      
2013-12-30  NA      
2013-12-30  22      
2013-12-30  24
2013-12-30  25

What I want to do?

I want to calculate mean and median of the coins column, ignoring missing values.

What i have done so far?

1. Grouped the data on date variable by_date <- group_by(df, date)
2. Summarised data using:by_date %>% summarise_each_(funs(mean(., na.rm = TRUE), median(., na.rm=TRUE)), names(by_date)[2])

Question The results returned by summarise_each_ show NaN for date 2013-10-01. Does that mean the function is not ignoring missing values?

Answer 1

The problem here is that all the values for 2013-10-01 are NA, so there can't be a mean. The NaN is R trying to tell you this.

If you'd rather just not have 2013-10-01 show up in the summary, one option is to get rid of NA values upfront like this:

by_date<-group_by(df[!is.na(df$coins),],date)