The repo includes the following:
- xgb_model.pckl :XGboost model for classifying satisfactory and unsatisfactory chatbot-customer conversations
- ISODS_train.ipynb:Code for model training
- ISODS_test.ipynb:Code to load the model and predict on test data
- vietnamese-stopwords.txt: Text file containing list of Vietnamese stop words
All the files are required to be in the same folder along with training and test csv files
- pandas (v1.1.3)
- numpy (v1.19.2)
- regex (v2.2.1)
- scikit-learn (v0.23.2)
- pyvi
If using a jupyter notebook, it is convenient to create a separate conda environment and activating it prior installing these packages.
conda create --name <env_name>conda activate <env_name>Use the conda manager to install the packages (except pyvi, which is installed through pip)
Competition result: 2nd