XGBoost Multiclass & Multilabel

Here are the examples for XGboost multiclass and multilabel classification cited in the Medium article I wrote.

Multiclass classification tips

For multiclass, you want to set the objective parameter to multi:softmax.

objective: multi:softmax: set XGBoost to do multiclass classification using the softmax objective, you also need to set num_class(number of classes)

Multiclass examples in xgboost-multiclass/

Requirements

Install dependencies by running:

pip install -r requirements.txt

(You want to be using an environment to install this dependencies. If you're unsure on how to use one, follow the docs.)

Datasets

[1] Wine Data Set: does not need to be downloaded. Can be loaded from Sklearn module using

from sklearn.datasets import load_wine

[2] Anuran Calls (MFCCs) Data Set

Download the zip folder to datasets/.

wget https://archive.ics.uci.edu/ml/machine-learning-databases/00406/Anuran%20Calls%20\(MFCCs\).zip -P datasets

Extract the zip folder so we can access Frogs_MFCCs.csv.

unzip datasets/Anuran\ Calls\ \(MFCCs\).zip -d datasets