This project has the aim to define a Sentiment Analysis on Yelp Dataset for review data classification. Furthermore, we make a fake review model using mainly python ML libraries.
- Introduction: It is dedicated to an introduction on the problem instance and general description of the notebook contents
- Loading Dataset: Description on how we loaded information on the pandas dataframe
- Data Analysis: Analysis on the data types and values distribution in the entire dataset
- Data Pre-processing: Manipulation of information to prepare data for model input
- Data modelling: Building and training of models
- Data results: Shows results on training phase and final metrics
- Conclusions: Final observations
We produces a requirement.txt to use in the setting that should mantain dependencies consistency between different deployment. If you try to deploy it and have some dependencies problem; please, open an issue in the repository.
Models are some samples about evaluation on some techniques in relation to the Yelp dataset. In other words, we can define a better model using some state-of-art architectures like Convolutional LSTM, biLSTM or some of their variation. These models are in the version 2 which evaluates performance in relations to the old ones.
- Mario Sessa (@kode-git)