How to perform modulo on DateTime or Time type in Julia?

user_13495049 0
Viewed 89

I have some temporal experimental data in a DataFrame. I would like to round the timestamps to the nearest 10 seconds and then drop any rows that are not on an even minute. I have everything but the last step working below.

df = DataFrame()
df.Time = ["11:05:02", "11:05:23", "11:05:34", "11:05:42", "11:06:01"]
df.Data = rand(5)
df.Time = DateTime.(df.Time, DateFormat("H:M:S")) # I would rather just use Time type
df.Time = round.(df.Time, Second(10))             # but round is only defined for DateTime type

julia> df
5×2 DataFrame
 Row │ Time                 Data     
     │ DateTime             Float64  
─────┼───────────────────────────────
   1 │ 0001-01-01T11:05:00  0.987827
   2 │ 0001-01-01T11:05:20  0.534373
   3 │ 0001-01-01T11:05:30  0.571214
   4 │ 0001-01-01T11:05:40  0.306041
   5 │ 0001-01-01T11:06:00  0.209411

julia> filter!(:Time => time -> time % Minute(1) == 0, df)
ERROR: MethodError: no method matching rem(::DateTime, ::Minute)
julia dataframe modulo time datetime
1 Answers

Regarding rounding you could always define a simple function such as:

function round10(t::Time)
    s= second(t) % 10
    s <=5 && return t - Second(s)
    return t + Second(10-s)
end

This will allow to store times as Time that is df.Time = Time.(df.Time, DateFormat("H:M:S")).

Selecting rows that are on an even minute after rounding could be accomplished as

julia> df[minute.(round10.(df.Time)) .% 2 .== 0, :]
1×2 DataFrame
│ Row │ Time     │ Data     │
│     │ Time     │ Float64  │
├─────┼──────────┼──────────┤
│ 1   │ 11:06:01 │ 0.537324 │

or

julia> filter!(:Time => t -> minute(round10(t)) % 2 == 0, df)
1×2 DataFrame
│ Row │ Time     │ Data     │ 
│     │ Time     │ Float64  │ 
├─────┼──────────┼──────────┤
│ 1   │ 11:06:01 │ 0.537324 │
user_99577100
Related