# Count cumulative number of unique elements

**URL:** <https://discourse.julialang.org/t/count-cumulative-number-of-unique-elements/42736>\
**Category:** Data\
**Tags:** query\
**Created:** [July 8, 2020, 3:01pm UTC](https://discourse.julialang.org/t/count-cumulative-number-of-unique-elements/42736 "2020-07-08T15:01:37Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![tpoisot](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/tpoisot/32/19812_2.png) [@tpoisot](https://discourse.julialang.org/u/tpoisot)\
**Post date:** [July 8, 2020, 3:01pm UTC](https://discourse.julialang.org/t/count-cumulative-number-of-unique-elements/42736/1 "2020-07-08T15:01:37Z")

</div>

I am trying to produce a species discovery curve, which is basically the number of unique species observed over time. So, at time t, the value is `length(unique(_.species_name))` for all observations up to time t - this in itself is super easy to do, but I am looking for a way to do it using `Query.jl`.

Is there a mechanism to iterate? The issue I had with `@groupby` is that it only gives access to a single value when grouping by date, and I’m looking for a way to group by “all values lower than the current one”.

I realize that this isn’t necessarily a problem to solve with `Query.jl` but I’m trying “for fun”.

---

<div class="post-metadata">

**Author:** ![pdeffebach](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/pdeffebach/32/10320_2.png) [@pdeffebach](https://discourse.julialang.org/u/pdeffebach)\
**Post date:** [July 8, 2020, 3:10pm UTC](https://discourse.julialang.org/t/count-cumulative-number-of-unique-elements/42736/2 "2020-07-08T15:10:45Z")

</div>

I don’t think there is anythng built-in here. But here is a function you can use

```julia
julia> function get_unique_cumulative(val)
       # assumes sorted by time
       s = Set{eltype(val)}()
       unique_across_time = Vector{Int}(undef, length(val))
       t = 0
       for i in 1:length(val)
           v = val[i]
           if v in s
              unique_across_time[i] = t
           else
              t += 1
              unique_across_time[i] = t
              push!(s, v)
           end
       end
       return unique_across_time
       end

```

in base dataframes you can do

```julia
julia> df = DataFrame(time = 1:25, val = rand(1:25, 25));

julia> transform(df, "val" => get_unique_cumulative => "cumulative_unique_species");

```