Support Vector Machine-2

In this video, we jump straight into R to build and deploy high-performance Support Vector Machine (SVM) classifiers for text data. We move from theory to implementation by exploring two vital ecosystem approaches for text mining.

What we cover in this hands-on coding tutorial:

Fast Linear SVMs with textmodel_svmlin: We dive into the quanteda.textmodels package to build a fast, linear SVM classifier. You will see how to fit the model directly on a sparse Document-Feature Matrix (DFM) and use the predict() function to classify brand-new, unseen text strings.

Deep Feature Extraction: Using the classic e1071 package, we look past simple prediction accuracy to explore model interpretability. We extract the internal feature weights to isolate and rank the top 20 most influential words driving the classification split for each target group or political party.

By the end of this session, you will know exactly how to train an SVM, project new textual data into your decision space, and reverse-engineer the model to pull out the definitive keywords shaping your predictions.

R-code: SVM
https://drive.google.com/file/d/1nQFg-vB7c90qSeKoLJ9bPH6UrLq6ra-M/view?usp=sharing
SVM PDF:
https://drive.google.com/file/d/1Mf1aL_TPyNVCwHhpDYZlzgqFW_A4Vubi/view?usp=sharing

Видео Support Vector Machine-2 канала Neeraj Kaushik