SA Ambulance Dashboard

1. Context

This is a public data dashboard experiment around SA Ambulance callout data.

The work started from a simple question: how can a public dataset be ingested, checked, and presented in a way that makes patterns easier to follow over time?

It is not official SA Ambulance Service communication.

2. The Problem

Public datasets are often useful but hard to read in their raw form.

The original workflow had the usual data problems:

• data had to be collected regularly
• raw messages needed cleaning and structuring
• dataset changes needed to be noticed
• summaries needed to be clearer than a raw table
• visual patterns over time needed to be easier to inspect

3. Constraints

• Source data can be inconsistent.
• The dashboard has to make uncertainty and dataset quality visible.
• The presentation needs to be readable for public-interest use.
• Automated ingestion should reduce manual update work.
• The project must avoid looking like an official service channel.

4. The System I Designed

A public-facing dashboard workflow with ingestion, checks, and clearer summaries.

Components

• automated data ingestion
• parsing and dataset structuring
• dataset checks and confidence notes
• public-facing summaries
• time-based charts and comparisons
• clearer visual presentation of patterns over time

How It Works

1. The ingestion process collects public callout data.
2. Raw messages are parsed into a structured dataset.
3. Checks look for missing values, format changes, and other quality issues.
4. Cleaned data is prepared for public-facing summaries.
5. The dashboard presents patterns over time in a clearer interface.

5. Before vs After

Before	After
Raw messages and scattered checks	Structured dataset and repeatable checks
Hard-to-scan tables	Public summaries and clearer charts
Manual confidence checks	Dataset health notes built into the workflow
Patterns hidden in rows	Time-based visual presentation

6. Stack & Architecture

• Python for ingestion and data preparation
• Selenium for source collection where required
• Pandas for cleaning and structuring
• Dataset checks for quality and change detection
• Power BI for dashboard presentation

The important part is not the charting tool. It is the repeatable path from public source data to a checked, readable dashboard.

7. What This Shows

This project shows how I think about public-interest data systems:

• automate the boring data collection work
• make dataset quality visible
• keep public summaries readable
• present patterns without pretending the dashboard is an official source

Project links

• See it live: View the SA Ambulance Dashboard
• Contact: Talk to Blake about data dashboard work

8. What I Would Build Next

• clearer public notes around data limits
• better summary pages for common callout categories
• stronger changelog around dataset and dashboard changes
• improved visual checks for long-term patterns