How to Start?

• Download yelp dataset
• Steps 1-4 must be run for each dataset.

1. Make Total Datset

• run 'preprocess_file(make_splited_datasets).ipynb'
• make 7 splited datsets(if splited dataset have above 30,000 rows, do sampling(30,000))

2. Make Summary about 'text' column

• run 'make_summary.ipynb'
• cuz, sentiment analysis model has max length about input sequence

3. Make Sentiment score about 'summary' column

• run 'make_sentiment_score.ipynb'
• based on our Positive/Negative Scale Conversion Algorithm

4. Make ensemble model

• run 'make_ensemble.ipynb'
• Finally, we make final dataset that convert 'text' column to 'sentiment_score'
• Thus, input vector was created for the ML model.
• Ensemble model generation for 7 datasets

5. Use created Model

• run 'pycaret_model_load_and_prediction'
• After make ensemble model, you can load model and use.

6. Make webpage using gradio

• run 'webpage.ipynb'
• The final model is available on the web page.