I trained a multi-output linear regression model that predicts Ethiopian Facebook users reaction to a political news post
I scraped facebook using fbcrawl (https://github.com/rugantio/fbcrawl)
I scraped data from a Facebook Ethiopian news pages
I used googletrans a free unofficial google trainslate library to convert from Amahric to English
the data that is use to train this model might be biased you can see the data i used to train this model in /data and the unprocessed data in /dataset
python 3.x
git clone https://github.com/robelkebede44/repol.git
pip install -r requirements.txtcd ./repol
python repol.py --text="Prime Minster Dr. Abiy Ahmed, launched the ambitious Green Legacy campaign that set a milestone to plant 200 million tree seedlings"
Pull request are welcome. specially new data crawlerd from diffrient Facebook news pages and please open an issue first