smohan22 / fresh_dash

FreshDash

Using Yelp reviews to optimize restaurant inspection

Data Sources:

• Seattle Yelp Data
• Seattle Food Establishment Inspection Data

Code Structure

/Data

• /features: features tables.
• /Seattle: put raw Yelp review, merged inspection instances, and inspection record data here.

/Data Exploration

• Link inspection record
• Data cleaning
• Exploratory analysis and visualization

/Feature Generation

• /Ngram: code and output

• /Topic Modeling:

  • code

  • visualization demo in Jupyter Notebook

  • mallet: to replicate labeled-LDA result, 1. use R code to generate rev_tm_violation_rating.txt. 2. run command below in the terminal:

  • bin/mallet import-file --input rev_tm_violation_rating.txt --output yelp-short.seq --stoplist-file yelp.stops --label-as-features --keep-sequence --line-regex '([^\t]+)\t([^\t]+)\t(.*)'
    
    bin/mallet run cc.mallet.topics.LabeledLDA --input yelp-short.seq --output-topic-keys yelp-llda.keys --output-doc-topics docsAsTopicsProbs.txt
    

    the output is rev_tm_violation_rating.txt

/Machine Learning Pipeline

• /config: machine learning model training configuration yaml files

• /mloutput: model evaluation output

• classifiers.py: model training

  • python3 classifiers.py <config file path>
    

• MagicLoop.ipynb: run classifiers.py in Jupyter Notebook for handy visualization and model comparison

Contributors

Ran Bi, Shambhavi Mohan, Minjia Zhu @Uchicago Harris, CAPP