Github Challenge: AI-Powered Property Price Chatbot in Under 4Hrs

alberts su — Tue, 14 Jan 2025 14:20:06 +0000

This is a submission for the GitHub Copilot Challenge : New Beginnings

What I Built

A Vue web app that incorporates Github Model (ChatGPT-4o mini) to analyze and answer questions about NSW (New South Wales, Australia) property data

Home ownership is a milestone in everyone's journey. Transitioning from a renter into a home owner represents a new beginning into the next stage in life!

This project was created to address the pain in buying the first property. Most often, as a first home buyer, we have no knowledge of which suburb to buy or what is a good price for a property. With the help of AI, we can ask all the questions that we might have!

Demo

https://albertsundjaja.github.io/nsw_property_price/

Repo

albertsundjaja / nsw_property_price

A simple webapp that can help analyze NSW property prices

NSW Property Price

This simple web app is an AI tool to help analyze NSW property price based on data provided by NSW Valuer General

This project make use of OpenAI's GPT-4o Mini provided by Github Models

Project Setup

if you are running locally, you will need to install all the node packages as below

npm install

Compile and Hot-Reload for Development

npm run dev

a sample data is provided in raw_data folder, however to generate a new one you will need to run the below command

npm run init

View on GitHub

Copilot Experience

This project made extensive use of Copilot help with autocomplete, edits, prompts and chat.

Autocomplete makes writing codes faster and easier with the suggested code as I type
Edit is pretty useful to create a scaffold for a function that I can modify to suit my specific needs
Prompt was used to generate the download.py initial code and provide guidance on how to host my Vue JS app in github. The data cleaning / aggregation codes were mostly AI generated!
Chat is very useful for debugging. By entering the error logs, Copilot was able to suggest what went wrong and how to fix them. I only have experience with Vue 2, but through chatting with Copilot, I was able to use Vue 3 and translate my old Vue coding knowledge to use the new Composition API provided by Vue 3. I was also able to integrate Vuetify into a vanilla Vite generated code without having to consult its documentation. Truly a time saver!

GitHub Models

This project make use of OpenAI's GPT-4o Mini provided by Github Models to analyze the NSW property prices processed data

However, model limitations of only able to take 8000 tokens input would make the response to be inaccurate as not all the data can be fed into the model for context. Only top 100 suburbs are given to the model because of reason above in each prompt request

Data is downloaded from NSW Valuer General

Conclusion

I have never imagined that I can make a working project in under 4 hours. AI is truly a game-changer in increasing software developers' productivity.
There is a lot to improve in this project, AI response to user's prompt will only be as accurate as the data fed into it as such it is important to ensure accurate and complete data is given to the model.

The project should help every first home buyers to take a better decision in one of their biggest investment in life. Buying the correct property which is suitable with the individual's goal and investment objective is key to achieve financial goals. No one wants to buy a value-declining house! :)

DEV Community: alberts su