Housol Old

Landing
insights
using vector search

Inspiration

The inspiration for Housol stemmed from the desire to empower homebuyers with data-driven insights in a challenging housing market. With rising property prices and economic uncertainty, many prospective buyers, especially millennials and Gen Z, struggle to make informed decisions. Housol was created to simplify this process, providing accurate forecasts and personalized recommendations.

What It Does

Housol leverages AI and historical data to predict future house prices, helping users determine when and where they can afford to buy a home. It provides users with budget-based and timeline-based predictions, comparing their financial situation with historical trends to offer actionable insights. Additionally, the platform showcases top housing options from the past three years and suggests comparable properties based on user preferences.

How We Built It

We began by mapping out the user flow and identifying key needs. The project evolved through design thinking, UI/UX design, data analysis, and development stages. We utilized TiDB's vector search to handle searching through a hugh dataset for housing data similarity as well as generating suggestions for users. The prediction models were built using historical housing data, with SARIMA models to forecast future trends based on initial similar data retrieved through TiDB vector search capability and using vector search to find top properties based on historical data.

Challenges We Ran Into

One of the main challenges involved using vector types within TiDB, particularly in setting up types in Python. The constraint of not being able to add a vector type after the table creation required very deliberate initial database design, emphasizing the importance of thoughtful schema planning from the outset. We also encountered challenges in loading our initial data into the system to help train the model for prediction.

Accomplishments That We're Proud Of

We are proud of building a platform that simplifies the home-buying planning process by providing users with personalized, data-driven insights that let them accurately plan and save towards a home. The successful integration of TiDB's vector search and the development of accurate prediction models are significant achievements.

What We Learned

The development process underscored the importance of flexible and scalable design, particularly when working with advanced database solutions like TiDB. We also learned how critical it is to plan database schemas carefully when dealing with vector types, ensuring that the system can scale and evolve over time without significant rework.

What's Next for Housol

Looking ahead, we plan to enhance Housol by incorporating more data sources, refining our prediction models, and expanding the platform's capabilities. We aim to offer personalized insights for sellers as well and potentially make this platform publicly available.