Build a Data Application with Streamlit

This repository provides the supporting code for my presentation entitled Build a Data Application with Streamlit.

I have also made a YouTube playlist of videos in case you would prefer the content in approximately 20-minute chunks.

Branching

This repository includes a series of branches, starting with 01-intro and ending with master. The idea is to traverse each branch in order. This will allow you to see the Chicago Parking Tickets application as we build it. Each branch represents the completed code for that section.

Running the Code

If you are running things locally, follow the relevant cheat sheet in the Cheat Sheets folder.

Requirements

This

Secrets

The code in this repository relies on a file that you'll have to create in code\.streamlit\secrets.toml. Note that you will need to create the .streamlit directory as well as the file. This file will contain all of the secrets that you'll need to follow along. Here is a sample of how it should look:

[db]
connection_string="DRIVER={ODBC Driver 17 for SQL Server};SERVER=localhost;DATABASE=ChicagoParkingTickets;Trusted_Connection=yes"
# Replace this connection string with whatever is appropriate for your environment.

[aoai]
endpoint = "YOUR ENDPOINT"
key = "YOUR KEY"
deployment_name = "YOUR DEPLOYMENT NAME"

[search]
endpoint = "YOUR ENDPOINT"
key = "YOUR KEY"
index_name = "YOUR INDEX NAME"

[speech]
key = "YOUR KEY"
region = "YOUR REGION"

Parking Tickets Analysis

To run the parking ticket analysis application, you will need SQL Server and a copy of the Chicago Parking Tickets database. You may also need to update the SQL Server connection string in code\.streamlit\secrets.toml if you are not running SQL Server using Windows authentication on localhost.

Azure OpenAI Integration

In order to try out the Azure OpenAI integration, you will need the following resources:

1. An Azure OpenAI endpoint. You will want to copy the URL endpoint for Azure OpenAI, as well as one of your access keys.
2. Create a GPT-4 deployment named something like "gpt-4"
3. Azure AI Search
4. Azure AI Speech

Note that Azure OpenAI no longer requires explicit approval for using the service in the scenarios we cover. You should not need explicit approval in order to use Azure OpenAI on your Azure subscription.