Cost-sensitive Boosting Pruning Trees for depression detection on Twitter

The Cost-senstive boosting pruning trees code should be compiled with scikit-learn 0.18.2

"""

@authors: Lei Tong

@Email: lt228@leicester.ac.uk

"""

This method is built upon sklearn_tree_post_prune and scikit-learn 0.18.X.

Usage

Step 0: Configure the Python Environment

conda create -n py36 python=3.6.8

Step 1: Install Required Dependencies

source activate py36
pip install -r requirements.txt

Step 2: Download the source code of scikit-learn 0.18.0 and extract it.

Step 3: Build Cython Extensions

cd ./sklearn_tree_post_prune/src/
cython _tree_prune.pyx
python setup.py build

Step 4: Copy the following files to the directory scikit-learn-0.18.X/sklearn/tree/

cp ./sklearn_tree_post_prune/src/build/lib_Completement with the system version/tree/_tree_prune.cpython-36m-darwin.so to scikit-learn-0.18.X/sklearn/tree/
cp ./tree_prune.py to scikit-learn-0.18.X/sklearn/tree/

Step 5: Copy the following files to the directory scikit-learn-0.18.X/sklearn/ensemble/ and add 'from . import boost_utils' to ensemble/__init__.py file:

cp ./boost_modify.py to scikit-learn-0.18.X/sklearn/ensemble/
cp ./boost_utils.py to scikit-learn-0.18.X/sklearn/ensemble/

Step 6: Change the directory to the root folder of scikit-learn-0.18.X and execute the following command:

 cd ./scikit-learn-0.18.X/
 pip install -e .

Testing

Open an IPython shell::

>>> from sklearn.datasets import make_classification 
>>> from sklearn.ensemble import CBPT 

>>> X,y = make_classification(n_samples=1000)
>>> clf = CBPT(n_estimators=100,penalty_coef=0.5,penalty_low_limit=1,learning_rate=0.5) 
>>> clf.fit(X,y) 
>>> clf.score(X,y)

Twitter Depression Detection Datasets

Tsinghua Twitter Depression Dataset: http://depressiondetection.droppages.com/

CLPsych 2015 Twitter Dataset: http://www.cs.jhu.edu/mdredze/clpsych-2015-shared-task-evaluation/

Acknowledgement

Many thanks to the authors of sklearn_tree_post_prune.

Citation

If you find CBPT useful in your research, please consider citing:

@article{tong2022cost,
  title={Cost-sensitive Boosting Pruning Trees for depression detection on Twitter},
  author={Tong, Lei and Liu, Zhihua and Jiang, Zheheng and Zhou, Feixiang and Chen, Long and Lyu, Jialin and Zhang, Xiangrong and Zhang, Qianni and Sadka, 	     Abdul and Wang, Yinhai and others},
  journal={IEEE Transactions on Affective Computing},
  year={2022},
  publisher={IEEE}
}