All projects
Machine learning · Notebook experiments
Reading the sentiment
Two related explorations of sentiment classification: a recurrent model for IMDB movie reviews and a bidirectional LSTM pipeline for tweets. The notebooks and accompanying write up cover preprocessing, tokenization, model training, and evaluation.
Engineering focus
Turning raw text into model ready sequences, comparing classification approaches, and inspecting the limitations of the pipeline.
Tools
Python, TensorFlow / Keras, pandas, and text preprocessing libraries.
Scope
Notebook experiments, not a production analytics service. The portfolio does not claim independently reproduced accuracy.
Next project
A more connected desktop