GPT-4 as Data Analyst

This repo contains the data and code for our paper "Is GPT-4 a Good Data Analyst?".

1. Requirements

1.1 OPENAI_API_KEY

Create an account and get the API key for OpenAI (https://openai.com).

OPENAI_API_KEY=YOUR_KEY

1.2 SERPAPI_KEY

Create an account and get the API key for google retrieval (https://serpapi.com).

SERPAPI_KEY=YOUR_KEY

1.3 Install requirements

pip install -r requirements.txt

1.4 Download Databases

Download the sql databases from "nvBench dataset" to the current directory and name it as nvBench-main.

2. Run the code

python main.py

3. Demo

We provide a demo and you can try it by running the following command:

cd demo/
python demo_main.py

You will first enter the OpenAI and SERPAPI keys according to the instructions:

* Please enter your OpenAI API key: 
* Please enter your Google API key:

Then you will be prompted to choose the sqlite file and sql schema:

* Please choose the database file to read: 
* Please choose the schema file to read:

You can select demo/apartment_rentals.sqlite and demo/schema.sql. Or you can play with your own data.

Next, you will be prompted to enter the question. Press enter to use the default question or enter your own question:

* Please enter your question (Press Enter to use the default question):

The pipeline starts to run now.

4. Data

The analysis results generated by GPT-4 and annotation results for each analysis bullet point are shown in data/analysis_results.xlxs.

Other data and experimental results will be released soon.

Citation

If the code is used in your research, please star our repo and cite our paper as follows:

@inproceedings{cheng2023gptda,
  title={Is GPT-4 a Good Data Analyst?},
  author={Liying Cheng and Xingxuan Li and Lidong Bing},
  booktitle={Findings of EMNLP},
  url={"https://arxiv.org/abs/2305.15038"},
  year={2023}
}