In this project, we aim to predict a business's stars rating based on all the review text for that business using neural network implementations in TensorFlow. Consider this problem as a regression problem.

(1) Report the RMSE and plot the lift chart of the BEST neural network model you have obtained. (2) Choose 3-5 arbitrary businesses from your test data (preferably from different categories).
Show the names, the true star ratings, and the predicted ratings (from your best model) of those businesses.

Dataset

https://www.yelp.com/dataset

The dataset contains several JSON files. You can find the format of the data here:
https://www.yelp.com/dataset/documentation/main

Requirements

• You are required to split data to training and test. Use training data to train your models and evaluate the model quality using test data.
• Use TF-IDF to extract features from reviews. If you experience low memory issue when using tfidfVectorzier, set parameters max_df, min_df, and max_features appropriately. • You must use EarlyStopping when training neural networks using Tensorflow.
• Tuning the following hyperparameters when training neural networks using Tensorflow and record how they affect performance in your report. Save all the models you have tried as a proof in your notebook and tabulate your findings in your report.
• Activation: relu, sigmoid, tanh • Number of layers and neuron count for each layer • Optimizer: adam and sgd.