In this project an extensive experiment has been done on sentiment analysis using diffrenet types of machine learning algorithms. This project also includes visualization of conclusions. There are two notebooks which are arranged as follows (This project is my final project of ML course in university)
In the first notebook i analyse algorithms on a big sentiment analysis dataset including 45,000 comments. Two vectorizatioj techniques have been applied to convert comments to vectors, these two are word2vec and bag of words. i also experiment these algorithms on 3 diffrent levels of text preprocessing. algorithms implemented in the first notebook are as follows
- Logistic Regression
- SVM
- KNN
- Neural Network
In the second notebook some other analysis has been done. i applied some unsupervised algorithms to cluster and visualize the 45,000 dataset. in the last part of experiment i tried to apply trasfer learning from the best model on the larger dataset to a smaller dasaset (including 500 comments). algorithms implemented in the second notebook are as follows
- Kmeans
- Guassian Mixture Model
- Mini-Batch Kmeans
- PCA
- Neural Network