Predicting Types of Cyberbullying from Tweets

Summary

This project aims to build a predictive model by comparing accuracy of a machine learning model (random forest), deep learning algorithms (CNN, LSTM and Pretrained Transformer) to categorize tweets into cyber bullying classes and non-cyberbullying classes.

Steps to Reproducing the Project

Reproducing cleaning_notebook.ipynb

1. Create a repository on Github. See here on how to do it
2. Create a google account.
3. Navigate to google drive and create a folder named NLP.
4. Download data from kaggle on your PC and upload it in the newly created folder NLP.
5. Download the notebook cleaning_notebook.ipynb found here on your PC. Navigate to File>Upload notebook. Navigate to the location of the downloaded notebook cleaning_notebook.ipynb then upload it
6. Create a new cell in the begining of the notebook
7. Install the list of packages used to create the project by copying the code below on the cell in colab. The requirements.txt file can be found here : link. Download it from github to your local PC, then upload it to colab as a file.

!pip install -r requirements.txt

8. After copying the command, navigate to Runtime>Run all.

9. After all the cells in the notebook have been executed, download the two files clean_df.cv and 'train_data.csvon your local PC. Upload the two files to your createdNLP` folder in google drive.

10. On colab, Navigate to File>Save a copy on Github. A pop up appears. On the pop-up, select the name of the github repositiory you created earlier.

11. Add notebooks/ to the begining of file path. Then, click ok.This saves your file to github.

Reproducing eda.ipynb

1. Download the notebook eda.ipynb found here on your PC. Navigate to File>Upload notebook. Navigate to the location of the downloaded notebook eda.ipynb then upload it.

2. Create a new cell in the begining of the notebook

3. Install the list of packages used to create the project by copying the code below on the cell in colab. The requirements.txt file can be found here : link. Upload it to colab as a file again.

!pip install -r requirements.txt

4. After copying the command, navigate to Runtime>Run all.

5. On colab, Navigate to File>Save a copy on Github. A pop up appears. On the pop-up, select the name of the github repositiory you created earlier.

6. Add notebooks/ to the begining of file path. Then, click ok.This saves your file to github.

Reproducing CountVectorizer_random forest.ipynb

1. Download the notebook CountVectorizer_random forest.ipynb found here on your PC. Navigate to File>Upload notebook. Navigate to the location of the downloaded notebook CountVectorizer_random forest.ipynb then upload it.

2. Create a new cell in the begining of the notebook.

3. Install the list of packages used to create the project by copying the code below on the cell in colab. The requirements.txt file can be found here : link.Upload it to colab as a file again.

!pip install -r requirements.txt

4. After copying the command, navigate to Runtime>Run all.

5. On colab, Navigate to File>Save a copy on Github. A pop up appears. On the pop-up, select the name of the github repositiory you created earlier.

6. Add notebooks/ to the begining of file path. Then, click ok.This saves your file to github.

Reproducing the_multiclass_text_classification_pytorch.ipynb

1. Download the notebook the_multiclass_text_classification_pytorch.ipynb found here on your PC. Navigate to File>Upload notebook. Navigate to the location of the downloaded notebook the_multiclass_text_classification_pytorch.ipynb then upload it.

2. Navigate to Runtime>Change runtime type select GPU as the hardware accelerator

3. Create a new cell in the begining of the notebook.

4. Install the list of packages used to create the project by copying the code below on the cell in colab. The requirements.txt file can be found here : link.Upload it to colab as a file again.

!pip install -r requirements.txt

5. After copying the command, navigate to Runtime>Run all.

6. On colab, Navigate to File>Save a copy on Github. A pop up appears. On the pop-up, select the name of the github repositiory you created earlier.

7. Add notebooks/ to the begining of file path. Then, click ok.This saves your file to github.

Reproducing hugging_face.ipynb.ipynb

1. Download the notebook the_multiclass_text_classification_pytorch.ipynb found here on your PC. Navigate to File>Upload notebook. Navigate to the location of the downloaded notebook hugging_face.ipynb.ipynb then upload it.

2. Navigate to Runtime>Change runtime type select GPU as the hardware accelerator

3. Create a new cell in the begining of the notebook.

4. Install the list of packages used to create the project by copying the code below on the cell in colab. The requirements.txt file can be found here : link.Upload it to colab as a file again.

!pip install -r requirements.txt

5. After copying the command, navigate to Runtime>Run all.

6. On colab, Navigate to File>Save a copy on Github. A pop up appears. On the pop-up, select the name of the github repositiory you created earlier.

7. Add notebooks/ to the begining of file path. Then, click ok.This saves your file to github.