DS-GA 1003 project: Fake review detection

This is a fake review detection system we are creating for NYU's DS-GA 1003 final project. The dataset we're using comes from Yelp and is described here. The training, validation and test sets can also be found as .csv files on the Codalab page.

Getting the data

In order to run this modeling pipeline, you need to download the .csv files from Codalab and place them in the data/ directory first. Once that's done, run the following Jupyter notebooks, in order:

1. Kelsey/Pre_vectorization_feature_engineering.ipynb
2. Aren/FeatureTransformation.ipynb
3. Sid/ratings.ipynb
4. Kelsey/downsample.ipynb
5. Sid/vectorize-count.ipynb
6. Sid/concat-features.ipynb

Once the features have been concatenated, the data will be in the data/ folder.

At this point, if you want to run through our baseline model, you can do so in Sid/baseline-model.ipynb.

Collaborators:

Kelsey Markey

Aren Dakessian

Guido Petri