My very first Ainize project “Performance Comparison of Binary and Multi Class Text Classification Models With scikit-learn and TensorFlow” is now published. This project gives you an idea of which model to choose given that those models will behave differently based on the given text input data that you are going to train with. To give you more intuitive idea, various metrics are provided including: accuracy, RAM & CPU usage, training and prediction runtime, confusion matrix, and etc.
Following models were trained to classify binary dataset (disaster Tweets) and the multiclass dataset (news article categorization):
scikit-learnModels: Logistic Regression, Naive Bayes, SVM (Support Vector Machine), Gradient Boosting
TensorFlowModels: CNN (Convolutional Neural Network) Model, BERT Transformer Pre-Trained Model + CNN
More about this project: