{"id":8489,"date":"2023-12-18T08:13:02","date_gmt":"2023-12-18T16:13:02","guid":{"rendered":"https:\/\/live-cometml.pantheonsite.io\/?p=8489"},"modified":"2025-04-29T12:01:47","modified_gmt":"2025-04-29T12:01:47","slug":"llms-exploring-data-with-yolopandas-and-comet","status":"publish","type":"post","link":"https:\/\/www.comet.com\/site\/blog\/llms-exploring-data-with-yolopandas-and-comet\/","title":{"rendered":"LLMs: Exploring Data with YOLOPandas \ud83d\udc3c and Comet"},"content":{"rendered":"\n
\"yolopandas,
Image source: Unsplash<\/a><\/figcaption><\/figure>\n\n\n\n

Have you ever imagined how cool it would be to analyze, explore and visualize your data in Pandas without typing a single line of code by yourself?<\/p>\n\n\n\n

The world is changing so fast with the advent of LLMs (large language models). During my quest for more knowledge on LLMs and how to better leverage what this new tech has in store, I came across YOLOPandas<\/a>, a tool to interact with Pandas objects via LLMs and LangChain<\/a>.<\/p>\n\n\n\n

This article will dive into YOLOPandas, flex its abilities, and incorporate it with Comet.<\/p>\n\n\n\n

Overview of YOLOPandas<\/h2>\n\n\n\n

YOLOPandas is based on Langchain, a powerful library allowing abstractions over language models from different providers. It lets you specify commands with natural language and execute them directly on Pandas objects. You can preview the code before executing or set yolo=True<\/code> to execute the code straight from the LLM.<\/p>\n\n\n\n

Installation<\/h3>\n\n\n\n
pip install yolopandas<\/span><\/pre>\n\n\n\n
Once imported into our notebook, YOLOPandas adds an llm<\/code> accessor to Pandas DataFrames. We will consider two examples of how this works and better understand how to use it for maximum productivity.<\/p>\n\n\n\n
Note<\/strong>: We will need an OpenAI API key, which can be found on our account here<\/a>, then we will need to set it as an environment variable.<\/p>\n\n\n\n
import<\/span> os\nos.environ['OPENAI_API_KEY'<\/span>] = '*******'<\/span><\/span><\/pre>\n\n\n\n
First Example: Using a Dataset<\/h2>\n\n\n\n
We will use a movie rating dataset that I got from Kaggle. It contains data about different movies, their runtimes, IMDb ratings, where they were produced, and lots more.<\/p>\n\n\n\n