Cumulative sum on a TSFrame

[chiraganand]

cumsum(ts) does not work.

julia> cumsum(ts)
ERROR: MethodError: no method matching +(::TSFrame, ::TSFrame)
Closest candidates are:
  +(::Any, ::Any, ::Any, ::Any...) at ~/.julia/juliaup/julia-1.7.3+0.x64.linux.gnu/share/julia/base/operators.jl:655
  +(::ChainRulesCore.AbstractThunk, ::Any) at ~/.julia/packages/ChainRulesCore/RbX5a/src/tangent_arithmetic.jl:122
  +(::ChainRulesCore.Tangent{P}, ::P) where P at ~/.julia/packages/ChainRulesCore/RbX5a/src/tangent_arithmetic.jl:146
  ...
Stacktrace:
  [1] add_sum(x::TSFrame, y::TSFrame)
    @ Base ./reduce.jl:24
  [2] iterate
    @ ./iterators.jl:515 [inlined]
  [3] collect_to!
    @ ./array.jl:782 [inlined]
  [4] collect_to_with_first!
    @ ./array.jl:760 [inlined]
  [5] _collect(c::UnitRange{Int64}, itr::Base.Iterators.Accumulate{typeof(Base.add_sum), TSFrame, Base._InitialValue}, #unused#::Base.EltypeUnknown, isz::Base.HasLength)
    @ Base ./array.jl:754
  [6] collect
    @ ./array.jl:649 [inlined]
  [7] #accumulate#793
    @ ./accumulate.jl:294 [inlined]
  [8] accumulate
    @ ./accumulate.jl:292 [inlined]
  [9] cumsum(itr::TSFrame)
    @ Base ./accumulate.jl:151
 [10] top-level scope
    @ REPL[38]:1

[Prajna1999]

hi @chiraganand cumsum seems not be implemented in the package (?) . I can start implementing it. Can you please assign me this task?

[chiraganand]

@Prajna1999 You can assign the task to yourself and submit a PR. I can review it.

[Prajna1999]

@chiraganand Base.cumsum for vectors only works along one dimension. So we cannot directly use ts as an Array-like argument for TSFrame object. For instance for a random TSFrame object
`

julia> ts_rand
10×3 TSFrame with Int64 Index
 Index  x1        x2        cumulative_sum_row
 Int64  Float64   Float64   Float64
───────────────────────────────────────────────
     1  0.374538  0.374538            0.374538
     2  0.873534  0.873534            1.24807
     3  0.290125  0.290125            1.5382
     4  0.336841  0.336841            1.87504
     5  0.662271  0.662271            2.53731
     6  0.776336  0.776336            3.31364
     7  0.503406  0.503406            3.81705
     8  0.750177  0.750177            4.56723
     9  0.910337  0.910337            5.47756
    10  0.464067  0.464067    

`

To cumulatively sum row wise for x2 column, you need to call

`

julia> cumsum(ts_rand.x2)
10-element Vector{Float64}:
 0.37453777969575874
 1.2480721438971558
 1.5381970305596713
 1.8750378636195264
 2.537308550691223
 3.313644735994737
 3.8170502654496943
 4.567227632902407
 5.477564954520991
 5.941632119720697
`
 To add a new column to the existing TSFrame object, you can modify the underlying dataframe directly by running
`
julia> ts_rand.coredata[!,:cumulative_sum_row_2]=cumsum(ts_rand.x2)
10-element Vector{Float64}:
 0.37453777969575874
 1.2480721438971558
 1.5381970305596713
 1.8750378636195264
 2.537308550691223
 3.313644735994737
 3.8170502654496943
 4.567227632902407
 5.477564954520991
 5.941632119720697

julia> ts_rand
10×4 TSFrame with Int64 Index
 Index  x1        x2        cumulative_sum_row  cumulative_sum_row_2
 Int64  Float64   Float64   Float64             Float64
─────────────────────────────────────────────────────────────────────
     1  0.374538  0.374538            0.374538              0.374538
     2  0.873534  0.873534            1.24807               1.24807
     3  0.290125  0.290125            1.5382                1.5382
     4  0.336841  0.336841            1.87504               1.87504
     5  0.662271  0.662271            2.53731               2.53731
     6  0.776336  0.776336            3.31364               3.31364
     7  0.503406  0.503406            3.81705               3.81705
     8  0.750177  0.750177            4.56723               4.56723
     9  0.910337  0.910337            5.47756               5.47756
    10  0.464067  0.464067            5.94163               5.94163
`

do you want a custom cumsum for TSFrame that would sum across columns? IMO that would make things difficult than a simple for loop accumulator over the underlying dataframe.