Of course, this brings me back to another problem, which is replacing a
column of a dataframe with an Array without having the Array converted to a
DataArray or a NullableArray. I have a large CSV file in which over 99% of
the columns do not contain nulls. Because some of the columns contain
nulls I can't use CSV.read(..., nullable = false) but I do want to convert
those columns that do not contain nulls to Arrays.
If I try to replace the column as
mydf[1] = Array(mydf[1])
I get a DataArray in the release version of DataFrames. (I'm not sure what
happens in your nl/nullable branch).
The way I avoid this is right now is by working with mydf.columns directly
but I always caution others not to work directly with fields of composite
types and I should probably follow my own advice.
On Wednesday, September 14, 2016 at 4:07:21 PM UTC-5, Douglas Bates wrote:
>
> Thanks. I didn't try that because I thought that a NullableArray was an
> Array so that conversion would be a no-op. I should have tested it.
>
> On Wednesday, September 14, 2016 at 4:02:04 PM UTC-5, Milan Bouchet-Valat
> wrote:
>>
>> Le mercredi 14 septembre 2016 à 13:56 -0700, Douglas Bates a écrit :
>> > I remember someone (Milan?, Alex?) writing a short cal to convert a
>> > NullableArray to an Array but I can't seem to find it now. Can
>> > someone refresh my memory, please?
>> What do you mean exactly? This works AFAICT:
>> julia> convert(Array, NullableArray(1:3))
>> 3-element Array{Int64,1}:
>> 1
>> 2
>> 3
>>
>>
>> Regards
>>
>
--
You received this message because you are subscribed to the Google Groups
"julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
