Check if value is missing in DataFrame indexed by values (not by numerical indices)

General Usage
1

I have the following DataFrame:

julia> directories
247×133 DataFrame. Omitted printing of 127 columns
│ Row │ receiverId       │ 2018-03-05           │ 2018-03-06           │ 2018-03-07              │
├─────┼──────────────────┼──────────────────────┼──────────────────────┼─────────────────────────┼
│ 1   │ 001bc50940820093 │ floor7:east:pat7302  │ floor7:east:pat7302  │ floor7:east:pat7302     │
│ 2   │ 001bc509408200fe │ floor5:south:pat5111 │ floor5:south:pat5111 │ floor5:southICU:pat5111 │
│ 3   │ 001bc5094082008e │ floor5:east:pat5307  │ floor5:east:pat5307  │ floor5:east:pat5307     │
│ 4   │ 001bc50940820018 │ floor7:south:ns04    │ floor7:south:ns04    │ floor7:south:ns04       │
│ 5   │ 001bc5094082005c │ floor5:west:ns02     │ floor5:west:ns02     │ floor5:west:ns02        │
│ 6   │ 001bc509408200ae │ floor8:west:ns01     │ floor8:west:ns01     │ floor8:west:ns01        │
│ 7   │ 001bc50940820028 │ floor7:south:pat7110 │ floor7:south:pat7110 │ floor7:south:pat7110    │
│ 8   │ 001bc50940820094 │ floor7:north:pat7217 │ floor7:north:pat7217 │ floor7:north:pat7217    │
│ 9   │ 001bc509408200b7 │ floor9:east:pat9301  │ floor9:east:pat9301  │ floor9:east:pat9301     │
│ 10  │ 001bc509408200c6 │ floor9:east:pat9304  │ floor9:east:pat9304  │ floor9:east:pat9304     │
⋮
│ 237 │ 001bc509408200f6 │ missing              │ missing              │ missing                 │
│ 238 │ 001bc50940820116 │ missing              │ missing              │ missing                 │
│ 239 │ 001bc509408200a6 │ missing              │ missing              │ missing                 │
│ 240 │ 001bc509408101f8 │ missing              │ missing              │ missing                 │
│ 241 │ 001bc50940820118 │ missing              │ missing              │ missing                 │
│ 242 │ 001bc5094082008d │ missing              │ missing              │ missing                 │
│ 243 │ 001bc5094082011a │ missing              │ missing              │ missing                 │
│ 244 │ 001bc50940820099 │ missing              │ missing              │ missing                 │
│ 245 │ 001bc50940820113 │ missing              │ missing              │ missing                 │
│ 246 │ 001bc5094082010a │ missing              │ missing              │ missing                 │
│ 247 │ 001bc509408200f2 │ missing              │ missing              │ missing                 │

I want to check if the receiverId 001bc509408200f6 is missing for a given date. I index the DataFrame in the following way, for a given date (column):

julia> directories[directories[:receiverId] .== "001bc509408200f6", Symbol("2018-03-05")]
1-element Array{Union{Missing, String},1}:
 missing

Here, the type of the output is a 1-element Array{Union{Missing, String},1}. If I use ismissing, I get:

julia> ismissing(directories[directories[:receiverId] .== "001bc509408200f6", Symbol("2018-03-05")])
false

which is not the expected output (at least for me). If use use dot notation, since I am evaluating an array,

julia> ismissing.(directories[directories[:receiverId] .== "001bc509408200f6", Symbol("2018-03-05")])
1-element BitArray{1}:
 true

I get the expected output in an unexpected (again, for me) type. Therefore, I can’t for example use it as the condition for an if statement (at least I haven’t figured out how).

What’s the right way to check whether a value is missing when indexing a DataFrame with an array of Booleans? Note that if I index using numbers, e.g.

julia> ismissing(directories[237,2])
true

this just works. However, for the processing I’m working on, it’s easier to index by receiverId and date (column) values.

2

The issue is that DataFrames has no way of knowing that there is only one oversvation with df.receiverid == "001bc509408200f6", so it returns an array with one element rather than a scalar. To not have this behavior would lead to type instability.

One way to get a scalar return is by using findfirst, which will give you the index of the first row for which df.receiverid == "001bc509408200f6" is true.