如何在Julia的TimeSeries中增加价值

Question

我想在多个时间序列中添加值，但失败了

using TimeSeries
ta1 = TimeArray([Date(2015, 10, 01), Date(2015, 11, 01)], [15, 16])
ta2 = TimeArray([Date(2015, 11, 01), Date(2015, 12, 01)], [11, 3])
ta3 = TimeArray([Date(2015, 12, 01), Date(2016, 1, 01)], [1, 5])
# m12 = merge(+, ta1, ta2, ta3)

我希望m12应该等于

m12 == TimeArray([Date(2015, 10, 01), Date(2015, 11, 01),
                  Date(2015, 12, 01), Date(2016, 1, 1)],
                 [15,27,4,5])

错误消息

ERROR: MethodError: no method matching merge(::typeof(+), ::TimeArray{Int64,1,Date,Array{Int64,1}}, ::TimeArray{Int64,1,Date,Array{Int64,1}}, ::TimeArray{Int64,1,Date,Array{Int64,1}})
Closest candidates are:
  merge(::TimeArray{T,N,D,A} where A<:AbstractArray{T,N} where D<:TimeType where N, ::TimeArray{T,N,D,A} where A<:AbstractArray{T,N} where D<:TimeType where N, ::TimeArray{T,N,D,A} where A<:AbstractArray{T,N} where D<:TimeType where N, ::TimeArray{T,N,D,A} where A<:AbstractArray{T,N} where D<:TimeType where N...; kw...) where T at /home/dlin/.julia/packages/TimeSeries/8Z5Is/src/combine.jl:78
  merge(::Function, ::AbstractDict, ::AbstractDict...) at abstractdict.jl:314
  merge(::NamedTuple, ::Any) at namedtuple.jl:264
  ...
Stacktrace:
 [1] top-level scope at REPL[5]:1

Answer 1

[我没有看到在merge的vcat /map/ TimeSeries机器中轻松实现此目的的方法，如果可以将序列转换为DataFrame，我可能会这样做：

julia> df = join(DataFrame(ta1), DataFrame(ta2), DataFrame(ta3),  on = :timestamp, kind = :outer, makeunique=true)
4×4 DataFrame
│ Row │ timestamp  │ A       │ A_1     │ A_2     │
│     │ Date       │ Int64⍰  │ Int64⍰  │ Int64⍰  │
├─────┼────────────┼─────────┼─────────┼─────────┤
│ 1   │ 2015-10-01 │ 15      │ missing │ missing │
│ 2   │ 2015-11-01 │ 16      │ 11      │ missing │
│ 3   │ 2015-12-01 │ missing │ 3       │ 1       │
│ 4   │ 2016-01-01 │ missing │ missing │ 5       │

julia> df.A += (df.A_1 .+ df.A_2)
4-element Array{Int64,1}:
 15
 27
  4
  5

julia> TimeArray(df.timestamp, df.A)
4×1 TimeArray{Int64,1,Date,Array{Int64,1}} 2015-10-01 to 2016-01-01
│            │ A     │
├────────────┼───────┤
│ 2015-10-01 │ 15    │
│ 2015-11-01 │ 27    │
│ 2015-12-01 │ 4     │
│ 2016-01-01 │ 5     │