# Filter a dataframe based on a dictionary

**URL:** <https://discourse.julialang.org/t/filter-a-dataframe-based-on-a-dictionary/57738>\
**Category:** New to Julia\
**Tags:** question, package, dataframes\
**Created:** [March 22, 2021, 5:47pm UTC](https://discourse.julialang.org/t/filter-a-dataframe-based-on-a-dictionary/57738 "2021-03-22T17:47:26Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![mkarikom](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/mkarikom/32/7096_2.png) [@mkarikom](https://discourse.julialang.org/u/mkarikom)\
**Post date:** [March 22, 2021, 5:47pm UTC](https://discourse.julialang.org/t/filter-a-dataframe-based-on-a-dictionary/57738/1 "2021-03-22T17:47:26Z")

</div>

Is it possible to filter a data frame based on a dictionary whose keys are a subset of the columns and whose values match the dict?

For example, something like this (does not work):

```julia
df = DataFrame(a=[1,2,3],b=["a","b","c"],c=[1.,0,missing])
3×3 DataFrame
 Row │ a b c         
     │ Int64 String Float64?  
─────┼──────────────────────────
   1 │ 1 a 1.0
   2 │ 2 b 0.0
   3 │ 3 c missing   

fd = Dict(:a=>1,:b=>"a") # filter on this

filter(keys(fd) => (keys(fd)) .== values(fd),df) # try to filter on `fd`

ERROR: MethodError: no method matching getindex(::DataFrames.Index, ::Base.KeySet{Symbol,Dict{Symbol,Any}})

```

---

<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:** [March 22, 2021, 6:08pm UTC](https://discourse.julialang.org/t/filter-a-dataframe-based-on-a-dictionary/57738/2 "2021-03-22T18:08:54Z")

</div>

This should work

```julia
julia> filter(df) do row
       tokeep = true
       for k in keys(fd)
           if row[k] != fd[k]
               tokeep = false
           end
       end
       return tokeep
       end
1×3 DataFrame
 Row │ a b c
     │ Int64 String Float64?
─────┼─────────────────────────
   1 │ 1 a 1.0

```

Not sure how to make this using the `source => fun` syntax, though.

EDIT: Doing this with the `source => fun` syntax is very hard. But will be easier in DataFrames 1.0 (upcoming in a few months) which contains the function `subset`.

---

<div class="post-metadata">

**Author:** ![Derek\_Vetsch](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/derek_vetsch/32/8552_2.png) [@Derek\_Vetsch](https://discourse.julialang.org/u/Derek_Vetsch)\
**Post date:** [March 22, 2021, 6:10pm UTC](https://discourse.julialang.org/t/filter-a-dataframe-based-on-a-dictionary/57738/3 "2021-03-22T18:10:34Z")

</div>

`filter` takes arrays of columns as the left hand side of the pair in the first argument. The problem with what you have above is that `keys` returns `Base.KeySet` and `values` returns a `Base.ValueIterator`. If you use a generator that should work. Using your code above:

```julia
df = DataFrame(a = 1:3, b = ["a", "b", "c"], c = [1., 0, missing])
fd = Dict(:a => 1, :b => "a")
### Changes start here
filter([key for key in keys(fd)] => ((a, b) -> [a, b] == [val for val in values(fd)]), df)

```