Chat with information the straightforward approach in R or Python

Save the next code to app.py and launch the app with shiny run app.py. I ran it contained in the Positron IDE with the Shiny extension put in; VS Code ought to work the identical:

import chatlas
import querychat as qc
import pandas as pd
from shiny import App, render, ui
from dotenv import load_dotenv
from pathlib import Path

load_dotenv()

video games = pd.read_feather("games_py.feather")


def use_openai_models(system_prompt: str) -> chatlas.Chat:
   return chatlas.ChatOpenAI(
       mannequin="gpt-4.1",
       system_prompt=system_prompt,
   )  

querychat_config = qc.init(
    data_source=video games,
    table_name="video games",
    greeting=Path("greeting.md"),
    data_description=Path("data_dictionary_py.txt"),
    create_chat_callback= use_openai_models 
)

# Create UI
app_ui = ui.page_sidebar(
    qc.sidebar("chat"),
    ui.output_data_frame("data_table"),
)


# Shiny server logic
def server(enter, output, session):
    # This create a querychat object utilizing the configuration from above
    chat = qc.server("chat", querychat_config)

    #    chat.df() is the filtered/sorted reactive information body
    @render.data_frame
    def data_table():
        return chat.df()

# Create Shiny app
app = App(app_ui, server)

You need to now have a fundamental working Shiny app that solutions questions on NFL information.

There’s much more you are able to do with these packages, together with connecting to an SQL database and enhancing the Shiny software. Posit CTO Joe Cheng, the creator of querychat, demo’d an expanded Python model as a full-blown querychat dashboard app. There’s a template for it within the Shiny for Python gallery:

There’s additionally a demo repo for an R sidebot you’ll be able to run regionally by cloning the R sidebot GitHub repository. Or, you’ll be able to study the app.R file within the repository to get concepts for the way you may create one.