# Coercing Data Type of Large Matrix

**URL:** <https://discourse.julialang.org/t/coercing-data-type-of-large-matrix/89494>\
**Category:** Performance\
**Tags:** package, performance, machine-learning\
**Created:** [October 29, 2022, 4:02pm UTC](https://discourse.julialang.org/t/coercing-data-type-of-large-matrix/89494 "2022-10-29T16:02:00Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![nosewitz](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/nosewitz/32/39014_2.png) [@nosewitz](https://discourse.julialang.org/u/nosewitz)\
**Post date:** [October 29, 2022, 4:02pm UTC](https://discourse.julialang.org/t/coercing-data-type-of-large-matrix/89494/1 "2022-10-29T16:02:00Z")

</div>

I have a sparse matrix with dimensions 800 x 100,000 filled solely with binary data (1 or 0), when loading it into a table with MLJ, their type is determined to be count data. I am able to change it into the OrderedFactor type through the coerce function, but this is very slow. Is there any way of making the table interface automatically detect these as OrderedFactors instead of defaulting to Count? Any way of getting to this result would work, though that’s what I would imagine is the fastest.

Code:

```julia
function read_sparse(file::AbstractString, contains_labs = true)
    f1 = contains_labs == true ? 2 : 1
    ind = readlines(file)
    m = zeros(Int, lastindex(ind), 100_000)
    c = 1
    for i in ind
      i = replace(i, "\t" => " ") |> strip
      var = parse.(Int, split(i))
      for j in f1:lastindex(var)
        m[c, var[j]] = 1 
        
      end
      c += 1 
    end
    sparse(m)
end # -> returns a Sparse 800 x 100,000 matrix

```

```julia
train = read_sparse("train.txt") 
train_table = table(train)
X = coerce(train_table, Count => OrderedFactor); # This is the bottleneck

```