filter date from a string JULIA-CodePudding

I have the next format for my date in my dataframe:

Typeof(DateTime("2021-12-17T06:00:00"))
Feather.Arrow.Timestamp{Microsecond}

I want to filter the data by date, but I can't because of the type. I tried to chop it, but again because of the type I couldnt. MethodError: no method matching chop(::Feather.Arrow.Timestamp{Microsecond}; head=10, tail=2) Closest candidates are: chop(::AbstractString; head, tail) at strings/util.jl:184 So I try to change the type using parse but it is not allowed. In R I use filter and I have no problem. What can I do?

CodePudding user response：

Going with the title date from string, here's a way

julia> using Dates

julia> dt = DateTime("2021-12-17T06:00:00")
2021-12-17T06:00:00

julia> typeof(dt)
DateTime

julia> Dates.day(dt)
17

julia> Dates.monthabbr(dt)
"Dec"

# etc...

Or in a data frame

julia> df = DataFrame(date=["2021-12-17T06:00:00","2021-12-18T06:00:00","2021-12-19T06:00:00","2021-12-20T06:00:00","2021-12-21T06:00:00"])
5×1 DataFrame
 Row │ date                
     │ String              
─────┼─────────────────────
   1 │ 2021-12-17T06:00:00
   2 │ 2021-12-18T06:00:00
   3 │ 2021-12-19T06:00:00
   4 │ 2021-12-20T06:00:00
   5 │ 2021-12-21T06:00:00

julia> df.date = DateTime.(df[:,:date])

julia> df
5×1 DataFrame
 Row │ date                
     │ DateTime            
─────┼─────────────────────
   1 │ 2021-12-17T06:00:00
   2 │ 2021-12-18T06:00:00
   3 │ 2021-12-19T06:00:00
   4 │ 2021-12-20T06:00:00
   5 │ 2021-12-21T06:00:00

julia> Dates.day.(df.date)
5-element Vector{Int64}:
 17
 18
 19
 20
 21

CodePudding user response：

To filter a DataFrame's row by value, see the Subsetting section of the DataFrames manual.

julia> df = DataFrame(n = 1:16, dates = DateTime("2021-12-17T06:00:00"):Day(1):DateTime("2022-01-01T06:00:00"));

julia> summary(df)
"16×2 DataFrame"

julia> df[DateTime("2021-12-25") .<= df.dates .<= DateTime("2021-12-31"), :]
6×2 DataFrame
 Row │ n      dates               
     │ Int64  DateTime            
─────┼────────────────────────────
   1 │     9  2021-12-25T06:00:00
   2 │    10  2021-12-26T06:00:00
   3 │    11  2021-12-27T06:00:00
   4 │    12  2021-12-28T06:00:00
   5 │    13  2021-12-29T06:00:00
   6 │    14  2021-12-30T06:00:00

julia> #OR:

julia> datefilter(dates) = DateTime("2021-12-25") .<= dates .<= DateTime("2021-12-31")
datefilter (generic function with 1 method)

julia> subset(df, :dates => datefilter)
6×2 DataFrame
 Row │ n      dates               
     │ Int64  DateTime            
─────┼────────────────────────────
   1 │     9  2021-12-25T06:00:00
   2 │    10  2021-12-26T06:00:00
   3 │    11  2021-12-27T06:00:00
   4 │    12  2021-12-28T06:00:00
   5 │    13  2021-12-29T06:00:00
   6 │    14  2021-12-30T06:00:00

For the specifics of what you want to accomplish here, it's most useful if you can show us the actual code you've tried and what its intended purpose is. (chop in Julia is used to process strings. tidyr-type chop may be replaced by groupbys instead, but that again depends on what your end goal here is.)

CodePudding user response：

the problem I have is that my DataFrame looks like:

julia> df
5×1 DataFrame
 Row │ date                
     │ DateTime            
─────┼─────────────────────
   1 │ DateTime(2021-12-17T06:00:00)
   2 │ DateTime(2021-12-18T06:00:00)
   3 │ DateTime(2021-12-19T06:00:00)
   4 │ DateTime(2021-12-20T06:00:00)
   5 │ DateTime(2021-12-21T06:00:00)

Rather than:

 Row │ date                
     │ DateTime            
─────┼─────────────────────
   1 │ 2021-12-17T06:00:00
   2 │ 2021-12-18T06:00:00
   3 │ 2021-12-19T06:00:00
   4 │ 2021-12-20T06:00:00
   5 │ 2021-12-21T06:00:00

So, that´s why I can´t filter the data because of the type. The first has type: Vector{Feather.Arrow.Timestamp{Microsecond}} while the second has type : DateTime.