XGBoost for Node.js

This module provides the runtime for XGBoost model. Run existing XGBoost model with Node.js.

Require a C++ compiler to build XGBoost library.

Install

from npm

npm install xgboost

from GitHub

git clone --recursive git@github.com:nuanio/xgboost-node.git
npm install

Example

Train XGBoost model with Python, and then save the model file.

import xgboost as xgb
from sklearn import datasets

iris = datasets.load_iris()
dtrain = xgb.DMatrix(iris.data, label = iris.target)
param = {
    'max_depth': 3,
    'eta': 0.3,
    'objective': 'multi:softprob',
    'num_class': 3 # three class 0, 1, 2
}
num_round = 20
bst = xgb.train(param, dtrain, num_round)
bst.save_model("iris.xg.model")

Load the model with Node.js:

const xgboost = require('xgboost');
const model = xgboost.XGModel('iris.xg.model');

const input = new Float32Array([
  5.1,  3.5,  1.4,  0.2, // class 0
  6.6,  3. ,  4.4,  1.4, // class 1
  5.9,  3. ,  5.1,  1.8  // class 2
]);

const mat = new xgboost.matrix(input, 3, 4);
console.log(model.predict(mat));
// {
//   value: [
//     0.991, 0.005, 0.004, // class 0
//     0.004, 0.990, 0.006, // class 1
//     0.005, 0.035, 0.960, // class 2
//   ],
//   error: undefined,      // no error
// }

const errModel = xgboost.XGModel('data/empty');
console.log(errModel);
console.log(errModel.predict());

Checkout more examples in the test cases.

Functions

XGMatrix
matrix(data, row, col, missing = NaN) ⇒ XGMatrix
restoreMatrix(file) ⇒ XGMatrix
matrixFromCSC(data, indptr, indices, n = 0) ⇒ XGMatrix
matrixFromCSR(data, indptr, indices, n = 0) ⇒ XGMatrix
XGModel
XGModel(file) ⇒ XGModel
XGModel.predict(xgmatrix, mask = 0, ntree = 0) ⇒ Result
Result

XGMatrix

Kind: object - input matrix for XGModel

Field Type Description
matrix internal readonly property
error Error error status

matrix(data, row, col, missing) ⇒ XGMatrix

Kind: global function Returns: XGMatrix - xgboost matrix

Param Type Description
data Float32Array input matrix
row Integer matrix row
col Integer matrix col
missing Number = NaN missing value place holder 
const xgboost = require('xgboost');
const input = new Float32Array([
  5.1,  3.5,  1.4,  0.2, // class 0
  6.6,  3. ,  4.4,  1.4, // class 1
  5.9,  3. ,  5.1,  1.8  // class 2
]);
const mat = new xgboost.matrix(input, 3, 4);

XGMatrix.col() ⇒ Result

Kind: member function Returns: Result - return matrix column size

mat.col();

XGMatrix.row() ⇒ Result

Kind: member function Returns: Result - return matrix row size

mat.row();

restoreMatrix(file) ⇒ XGMatrix

Kind: global function Returns: XGMatrix - xgboost matrix

Param Type Description
file string input matrix file path 
const matFromFile = xgboost.restoreMatrix('test/data/xgmatrix.bin');

matrixFromCSC(data, indptr, indices, n) ⇒ XGMatrix

Kind: global function Returns: XGMatrix - xgboost matrix

Param Type Description
data Float32Array input matrix
indptr Uint32Array pointer to col headers
indices Uint32Array findex
n Integer = 0 number of rows; when it's set to 0, then guess from data 
// [
//   1, 2, 3, 1,
//   0, 1, 2, 3,
//   0, 1, 1, 1,
// ]
const sparseCSC = xgb.matrixFromCSC(
  new Float32Array([1, 2, 1, 1, 3, 2, 1, 1, 3, 1]),
  new Uint32Array([0, 1, 4, 7, 10]),
  new Uint32Array([0, 0, 1, 2, 0, 1, 2, 0, 1, 2]),
  0);

matrixFromCSR(data, indptr, indices, n) ⇒ XGMatrix

Kind: global function Returns: XGMatrix - xgboost matrix

Param Type Description
data Float32Array input matrix
indptr Uint32Array pointer to row headers
indices Uint32Array findex
n Integer = 0 number of columns; when it's set to 0, then guess from data 
// [
//   1, 2, 3, 1,
//   0, 1, 2, 3,
//   0, 1, 1, 1,
// ]
const sparseCSR = xgboost.matrixFromCSR(
  new Float32Array([1, 2, 3, 1, 1, 2, 3, 1, 1, 1]),
  new Uint32Array([0, 4, 7, 10]),
  new Uint32Array([0, 1, 2, 3, 1, 2, 3, 1, 2, 3]),
  0);

XGModel

Kind: object - Trained XGModel

Field Type Description
model internal readonly property
error Error error status

XGModel(file) ⇒ XGModel

Kind: global function Returns: XGModel - xgboost model

Param Type Description
file string model file path 
const model = xgboost.XGModel('test/data/iris.xg.model');

XGModel.predict(xgmatrix, mask = 0, ntree = 0) ⇒ Result

Kind: member function Returns: Result - prediction result with Float32Array

Param Type Description
matrix XGMatrix input matrix
mask Integer = 0 bit-mask of options taken in prediction, possible values,
0:normal prediction,
1:output margin instead of transformed value,
2:output leaf index of trees instead of leaf value, note leaf index is unique per tree,
4:output feature contributions to individual predictions
ntree Integer = 0 limit number of trees used for prediction,
this is only valid for boosted trees when the parameter is set to 0,
we will use all the trees		 
model.predict(mat);

Result

Kind: object

Field Type Description
value Float32Array | number prediction result or method result
error Error error status

Contributing

Thank you for your interest to contribute to the XGBoost Node.js module.

Please submit an issue to suggest new feature before submitting pull requests.