Final report peer review - dah384 #130
Description
The project is to predict the rating of Airbnb in California (emphasizing in LA area). The datasets are taken from insideairbnb.com with some physical features like room type, bed type as well as location (Zip code) and etc.
What I like:
- I like how they put case studies to help the reader understand how to really apply their model in real life.
- They put human factors to consideration in section 5 (Discussion on Results) which as their basis of prediction
- The future work seems to be reasonable and align with their project.
Some suggestions:
- The motivation of the whole project is not that clear. Stating it on the first section (Project Overview) will help the reader to get this straight away.
- They drop more than 50% of the data that doesnt fit the minimum threshold level. Maybe give a rationale of why the prediction would still be doing well given this can biased the result.
- They mentioned about performing some feature engineering tricks, but they do not elaborate what kind of tricks and impacts this will give to the whole predictions.
Overall a good project but the WMD discussion might not align with the real definition. WMD is created if it creates a sort of negative feedback loop that amplify the negative effect of that is caused by the prediction.