gigasandwich/giga-malagasy-sentiment-analysis

# Goal Fetch reviews of products from random sources to do sentimental analysis so `we may predict if a comment of a review is POSITIVE or NEGATIVE` in `MALAGASY` language # Steps - [x] Format the reviews in `data/original.txt` into a more raeadable format (CSV) ONLY with positive/negative reviews - [x] Load the CSV (to train the model) - [x] Train and test the model - [x] Traduct the language to Malagasy - [X] Train and test the model for Malagasy language - [] Do an e-commerce like website to add comments on a product # How to run ```bash py run.py ``` # Modules To install non native modules of python-3: ```bash pip install pandas sklearn gensim openpyxl ``` # Note Best combo so far: trained_models/randomforest-tfidf.pkl ``` Accuracy: 0.7104677060133631 Classification report: precision recall f1-score support -1 0.69 0.89 0.78 254 1 0.77 0.48 0.59 195 accuracy 0.71 449 macro avg 0.73 0.68 0.68 449 weighted avg 0.72 0.71 0.69 449 Ity no vokatra tsara indrindra novidiko: 1 Tena halako ilay izy, serivisy ratsy be!: -1 Milay izy izany: -1 Hividy hafa koa aho amin ny manaraka: 1 Vokatra ratsy indrindra novidiko hatrizay: -1 Omeko 10/10 izany, tena tsara: 1 Tsy tsara: -1 Tsara: 1 Vokatra ara-barotra tena tsara: 1 Aza mividy ity vokatra ity: -1 Nahoana ny olona no mivarotra an ity, ity no vokatra ratsy indrindra eto: -1 ``` # Model size comparison