Create Small Sentiment Analysis

With the CSV attached, implement a solution for sentiment classification of movie reviews. The solution should only consider the language reviews and should classify those reviews as either: Positive, Average, or Negative. 1) Hand-craft or automatically create features that you think would be useful to classify the reviews. 2) Store the model training experiments in a way that it is easy to revisit and see/compare their metrics. 3) Makes sure to include at least one unittest for one of your classes (preferably your model class). 4) For inference purposes, create a REST API with a "/classify" endpoint that would take a single review (i.e. corresponding to a row in the dataset), and perform inference (e.g., required data transformations and model prediction) and finally return the label: Positive, Average, or Negative..