Comparison of a dataframe and a number

[jannefiluren]

I want to know if 2 is within a subset of my dataframe.

df = DataFrame(a = [1,1], b=[2,2], c=[3,3])

# This doesn´t give the right answer, it returns false since it compares a dataframe to a number (I guess)

df[:,2:3] .== 2 

# This works as I want to, returns an array of true/false

convert(Array, df[:,2:3]) .== 2

To me, the result of the first operation seems like something people might make mistakes with without noticing it. And the second solution is a bit cumbersome. Should this behaviour be changed?

[nalimilan]

As I noted on Slack: Sounds like a bug. df[:,2:3] .== 2 should either throw an error or return a data frame of the same shape as the original but with Bool columns. Maybe it already behaves differently on Julia 0.7, but I get an error when trying to load DataFrames right now.

[Nosferican]

On Julia 0.7 is the same behavior. However, at least in my pipelines for StatsModels dataframes should always be converted to arrays in order to have Array like behavior. For a discussion on enabling Array behavior for AbstractDataFrames see #583.

[nalimilan]

Fixed in Julia 1.0:

julia> df = DataFrame(a = [1,1], b=[2,2], c=[3,3]);
julia> df[:,2:3] .== 2 
ERROR: MethodError: no method matching iterate(::DataFrame)

See also #1513.