# MethodError: no method matching DMatrix
**URL:** https://discourse.julialang.org/t/methoderror-no-method-matching-dmatrix/68202
**Category:** New to Julia
**Tags:** question, error
**Created:** [September 15, 2021, 9:55am UTC](https://discourse.julialang.org/t/methoderror-no-method-matching-dmatrix/68202 "2021-09-15T09:55:32Z")
**Posts on this page:** 3
**Page:** 1
### Author:  [@Billpete002](https://discourse.julialang.org/u/Billpete002)
#### Post date: [September 15, 2021, 9:55am UTC](https://discourse.julialang.org/t/methoderror-no-method-matching-dmatrix/68202/1 "2021-09-15T09:55:33Z")
Hi All,
After fixing my one-hot encoding issue I’ve split my code into test and train but I am struggling to get XGBoost’s DMatrix to run. All of my variables are numeric (with the exception of my label, which I did test as numeric to see if that was the issue and I got the same error. At least in R the label needed to be categorical when making a DMatrix).
```julia
#train and test set
function partitionTrainTest(data, at = 0.7)
n = nrow(data)
idx = shuffle(1:n)
train_idx = view(idx, 1:floor(Int, at*n))
test_idx = view(idx, (floor(Int, at*n)+1):n)
data[train_idx,:], data[test_idx,:]
end
train,test = partitionTrainTest(df, 0.7) # 70% train
train_X = train[:, Not(1)]
train_Y = train[:, 1]
test_X = test[:, Not(1)]
test_Y = test[:, 1]
#train_Y = DataFrame(label = train_Y)
#test_Y = DataFrame(label = test_Y)
dtrain = DMatrix(train_X , label = train_Y)
```
I’ve tried the label as a dataframe or an array neither work and provide the same error.
```julia
dtrain = DMatrix(train_X, label = train_Y)
ERROR: MethodError: no method matching DMatrix(::DataFrame; label=CategoricalArrays.CategoricalValue{Float64, UInt32}[0.0, 0.0, 1.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 0.0 … 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0, 1.0])
```
I assume I need to somehow turn the dataframe into a matrix but I am not sure exactly how…
---
### Author:  [@Billpete002](https://discourse.julialang.org/u/Billpete002)
#### Post date: [September 16, 2021, 7:47am UTC](https://discourse.julialang.org/t/methoderror-no-method-matching-dmatrix/68202/2 "2021-09-16T07:47:52Z")
Here I figured out how to turn this into a matrix but I still end up with an error:
```julia
train_X = Matrix{Union{Missing, Float32}}(train[:, Not(1)])
train_Y = convert(Array{Float32}, train[:, :Business_Type])
test_X = Matrix{Union{Missing, Float32}}(test[:, Not(1)])
test_Y = convert(Array{Float32}, test[:, :Business_Type])
# DMatrix
dtrain = DMatrix(train_X, label = train_Y)
```
And here is the error:
```julia
ERROR: MethodError: no method matching DMatrix(::Matrix{Union{Missing, Float64}}; label=Float32[1.0, 0.0, 1.0, 1.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0 … 0.0, 1.0, 0.0, 0.0, 1.0, 1.0, 1.0, 0.0, 1.0, 0.0])
Closest candidates are:
DMatrix(::Matrix{var"#s56"} where var"#s56"<:Real)
```
I’ve also tried Float32 and Float64 and neither work (same error)…
```julia
train_X = Array{Union{Missing, Float64}}(train[:, Not(1)]) #448193×34 Matrix{Union{Missing, Float64}}
train_Y = Array{Union{Missing, Float64}}(train[:, :Business_Type]) #448193-element Vector{Union{Missing, Float64}}
```
Error:
```julia
ERROR: MethodError: no method matching DMatrix(::Matrix{Union{Missing, Float64}}; label=Union{Missing, Float64}[0.0, 0.0, 0.0, 1.0, 1.0, 0.0, 0.0, 0.0, 1.0, 0.0 … 0.0, 1.0, 0.0, 1.0, 0.0, 1.0, 0.0, 1.0, 0.0, 0.0])Closest candidates are:
DMatrix(::Matrix{var"#s56"} where var"#s56"<:Real)
```
---
### Author:  [@Billpete002](https://discourse.julialang.org/u/Billpete002)
#### Post date: [September 16, 2021, 10:41am UTC](https://discourse.julialang.org/t/methoderror-no-method-matching-dmatrix/68202/3 "2021-09-16T10:41:14Z")
Finally solved it with digging into the DMatrix!
```julia
function DMatrix(data::Matrix{<:Real}, transposed::Bool = false, missing = NaN32;
kwargs...)
handle = nothing
if !transposed
handle = XGDMatrixCreateFromMat(convert(Matrix{Float32}, data),
convert(Float32, missing))
else
handle = XGDMatrixCreateFromMatT(convert(Matrix{Float32}, data),
convert(Float32, missing))
end
```
The key part in the underlying function is that it was looking for REAL instead of Float32 - which is not what I’ve seen in the few examples online such as:
```julia
testArray = convert(Array{Float32}, test[:, [:annual_usage, :bracket_pricing, :min_order_quantity, :quantity, :year, :month, :day, :dayWeek, :supplier]])
```
Which is from [Starter Code](https://www.kaggle.com/wacaxx/xgboost-in-julia-starter-code/comments)
which doesn’t work any longer!