The goal of this project is to predict the rating of a hotel stay from a review in natural language using an Multilayer Perceptron (Neural Network). There are two components: a Jupyter Notebook to train the model and a Streamlit or Flask App to use/show the model.
The notebook is pretty straight forward.
-
Load the data (which is included in the repo, so it's easier for anyone to use it),
-
Separate data into training and test set
-
Preprocess data (which I did using a bag-of-words approach)
-
Design, train and save the network
-
TODO: Use test set to assess model accuracy