# Change datatype for subset of DataFrame columns

**URL:** <https://discourse.julialang.org/t/change-datatype-for-subset-of-dataframe-columns/75767>\
**Category:** Data\
**Tags:** dataframes\
**Created:** [February 4, 2022, 12:42am UTC](https://discourse.julialang.org/t/change-datatype-for-subset-of-dataframe-columns/75767 "2022-02-04T00:42:23Z")\
**Posts on this page:** 7\
**Page:** 1

<div class="post-metadata">

**Author:** ![enzomar](https://avatars.discourse-cdn.com/v4/letter/e/b4bc9f/32.png) [@enzomar](https://discourse.julialang.org/u/enzomar)\
**Post date:** [February 4, 2022, 12:42am UTC](https://discourse.julialang.org/t/change-datatype-for-subset-of-dataframe-columns/75767/1 "2022-02-04T00:42:23Z")

</div>

Suppose I have this dataframe:  
df = DataFrame(A = Any[rand() for i in 1:10], B = [“a”,“b”,“c”,“d”,“e”,“f”,“g”,“h”,“i”,“j”], C = Any[rand() for i in 1:10])

I want to change the datatype of columns A and B programmatically (the actual dataframe has \>20 columns!).

So far I have identified the columns with datatype “Any” with:  
cols = names(prices, eltype.(eachcol(prices)) .== Any)

Now, how do I iterate through the columns to change the datatype?

(I searched everywhere and I only found ways to change a single column)

---

<div class="post-metadata">

**Author:** ![Henrique\_Becker](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/henrique_becker/32/15443_2.png) [@Henrique\_Becker](https://discourse.julialang.org/u/Henrique_Becker)\
**Post date:** [February 4, 2022, 12:57am UTC](https://discourse.julialang.org/t/change-datatype-for-subset-of-dataframe-columns/75767/2 "2022-02-04T00:57:38Z")

</div>

```julia
julia> using DataFrames
[Info: Precompiling DataFrames [a93c6f00-e57d-5684-b7b6-d8193f3e46c0]

julia> df = DataFrame(A = Any[rand() for i in 1:10], B = string.('a':'j'), C = Any[rand() for i in 1:10])
10×3 DataFrame
 Row │ A B C        
     │ Any String Any      
─────┼─────────────────────────────
   1 │ 0.906792 a 0.674568
   2 │ 0.410943 b 0.370189
   3 │ 0.184457 c 0.759076
   4 │ 0.889315 d 0.332498
   5 │ 0.70852 e 0.839533
   6 │ 0.0455212 f 0.27601
   7 │ 0.955367 g 0.295744
   8 │ 0.435959 h 0.16552
   9 │ 0.186456 i 0.156159
  10 │ 0.529786 j 0.717429

julia> for name in names(df); df[!, name] = identity.(df[!, name]); end

julia> df
10×3 DataFrame
 Row │ A B C        
     │ Float64 String Float64  
─────┼─────────────────────────────
   1 │ 0.906792 a 0.674568
   2 │ 0.410943 b 0.370189
   3 │ 0.184457 c 0.759076
   4 │ 0.889315 d 0.332498
   5 │ 0.70852 e 0.839533
   6 │ 0.0455212 f 0.27601
   7 │ 0.955367 g 0.295744
   8 │ 0.435959 h 0.16552
   9 │ 0.186456 i 0.156159
  10 │ 0.529786 j 0.717429

```

---

<div class="post-metadata">

**Author:** ![DataFrames](https://avatars.discourse-cdn.com/v4/letter/d/e19b73/32.png) [@DataFrames](https://discourse.julialang.org/u/DataFrames)\
**Post date:** [February 4, 2022, 1:47am UTC](https://discourse.julialang.org/t/change-datatype-for-subset-of-dataframe-columns/75767/3 "2022-02-04T01:47:50Z")

</div>

```julia
transform!(df, Cols(:) .=> ByRow(identity), renamecols = false)

```

---

<div class="post-metadata">

**Author:** ![xinchin](https://avatars.discourse-cdn.com/v4/letter/x/54ee81/32.png) [@xinchin](https://discourse.julialang.org/u/xinchin)\
**Post date:** [February 5, 2022, 12:17am UTC](https://discourse.julialang.org/t/change-datatype-for-subset-of-dataframe-columns/75767/4 "2022-02-05T00:17:57Z")

</div>

I think `identity.(df)` works too.

Edit: though the solution which uses `transform!` seems faster for larger dfs.

---

<div class="post-metadata">

**Author:** ![enzomar](https://avatars.discourse-cdn.com/v4/letter/e/b4bc9f/32.png) [@enzomar](https://discourse.julialang.org/u/enzomar)\
**Post date:** [February 6, 2022, 7:47pm UTC](https://discourse.julialang.org/t/change-datatype-for-subset-of-dataframe-columns/75767/5 "2022-02-06T19:47:13Z")

</div>

@DataFrames I market your answer as the “solution”. On the other hand if the dataframe has many columns and not all of them require typing “narrowing down”, wouldn’t this solution be sub-optimal (requiring to process all columns vs. the one that really need to be processed)?

Coming from R I’m surprised it is not possible to just pass a vector with the column names that require processing…

---

<div class="post-metadata">

**Author:** ![Henrique\_Becker](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/henrique_becker/32/15443_2.png) [@Henrique\_Becker](https://discourse.julialang.org/u/Henrique_Becker)\
**Post date:** [February 6, 2022, 8:03pm UTC](https://discourse.julialang.org/t/change-datatype-for-subset-of-dataframe-columns/75767/6 "2022-02-06T20:03:42Z")

</div>

What do you mean, it is not possible to simply replace `Cols(:)` by a `Vector` of column names?

```julia
julia> transform!(df, ["A", "C"] .=> ByRow(identity), renamecols = false)
10×3 DataFrame
 Row │ A B C        
     │ Float64 String Float64  
─────┼─────────────────────────────
   1 │ 0.104016 a 0.551771
   2 │ 0.609946 b 0.211459
   3 │ 0.136688 c 0.595575
   4 │ 0.217122 d 0.543388
   5 │ 0.916172 e 0.514883
   6 │ 0.655476 f 0.502752
   7 │ 0.261034 g 0.543343
   8 │ 0.0641058 h 0.765185
   9 │ 0.817648 i 0.414756
  10 │ 0.413503 j 0.381042

```

---

<div class="post-metadata">

**Author:** ![enzomar](https://avatars.discourse-cdn.com/v4/letter/e/b4bc9f/32.png) [@enzomar](https://discourse.julialang.org/u/enzomar)\
**Post date:** [February 6, 2022, 8:21pm UTC](https://discourse.julialang.org/t/change-datatype-for-subset-of-dataframe-columns/75767/7 "2022-02-06T20:21:45Z")

</div>

That’s great!  
I thought it wasn’t possible. Newby here 🙂