transform xgb.DMatrix to dataframe or matrix

user_9218849 0
Viewed 1511

I have a xgb.Dmatrix which I want to convert to dataframe or matrix format, I would like to know if it is possible to convert.

data(agaricus.train, package='xgboost')
train <- agaricus.train
dtrain <- xgb.DMatrix(train$data, label=train$label)
class(dtrain)

when I try to convert dtrain to matrix or dataframe I get an error, can someone suggest what can be done

enter image description here

I need the dtrain to be in matrix or dataframe format

xgboost dplyr tidyverse r
2 Answers

This seems to be a problem of exporting data to Cpp. Where xgboost authors did not implement a getter sugar. At least dimnames and dim could be retrieve. Implemented methods:

> methods(class="xgb.DMatrix")
[1] [          dim        dimnames   dimnames<- getinfo    print      setinfo
[8] slice
see '?methods' for accessing help and source code

colnames works too

Example :

> dd <- xgb.DMatrix(cbind(a = 1:10, b = 1:10))
> colnames(dd)
[1] "a" "b"
> dim(dd)
[1] 10  2
> dimnames(dd)
[[1]]
NULL

[[2]]
[1] "a" "b"
> slice(dd, 1L)
xgb.DMatrix  dim: 1 x 2  info: NA  colnames: no

This is part of code (xgb.DMatrix body) where a data is exported to Cpp:

if (length(data) > 1)
            stop("'data' has class 'character' and length ",
                length(data), ".\n  'data' accepts either a numeric matrix or a single filename.")
        handle <- .Call(XGDMatrixCreateFromFile_R, data, as.integer(silent))
    }
    else if (is.matrix(data)) {
        handle <- .Call(XGDMatrixCreateFromMat_R, data, missing)
        cnames <- colnames(data)
    }
    else if (inherits(data, "dgCMatrix")) {
        handle <- .Call(XGDMatrixCreateFromCSC_R, data@p, data@i,
            data@x, nrow(data))
        cnames <- colnames(data)
    }
    else {
        stop("xgb.DMatrix does not support construction from ",
            typeof(data))
    }
user_54425270

This is not possible , got response from lead maintainer of Xgboost library

user_92188490
Related