HouseMatchGPT

• 

  Home Page

• 

  Output

• 

  Search Page

• 

  Housing Dataset

Inspiration

HouseMatchGPT was inspired by the need to streamline and enhance property searches using advanced AI and natural language. We aimed to simplify finding the perfect home by integrating data-driven insights with natural language processing.

What it does

HouseMatchGPT analyzes property data by incorporating features like school ratings and park proximity. It uses spaCy for intelligent keyword extraction and the OpenAI API to generate a refined table of the top 5 property options with detailed descriptions, making property searches smarter and more precise.

How we built it

We utilized the Boston housing dataset, applied feature engineering to add metrics like school ratings and distance to parks, and standardized the data with MinMaxScaler. Keywords were extracted using spaCy's en_core_web_sm model, and the OpenAI API was used to match and generate detailed property options. UI is created using streamlit.

Challenges we ran into

We faced challenges in integrating diverse data sources and ensuring the accuracy of our feature engineering. Additionally, refining keyword extraction to effectively match user queries and generate relevant property recommendations required extensive fine-tuning.

Accomplishments that we're proud of

We successfully combined multiple AI technologies to enhance property search capabilities, providing a user-friendly interface with AI-generated descriptions. Our system effectively refines search results based on detailed and dynamic criteria.

What we learned

We learned the importance of effective feature engineering and dynamic keyword extraction in improving search results. Integrating different AI technologies can significantly enhance the relevance and precision of recommendations.

What's next for HouseMatchGPT

We plan to expand HouseMatchGPT by incorporating additional features and data sources to further refine property recommendations. Future updates will include more advanced NLP techniques and broader geographic coverage to enhance user experience and accuracy.

Built With

• autogen
• llm
• natural-language-processing
• openai
• pandas
• python
• rag
• spacy
• streamlit