AI based housing marketplace to fuel real estate
This post is a short overview of an Abto Software AI project.
In the scope of our project, primarily focused around utilizing artificial intelligence for a modern marketplace, our company has cooperated with a US-based startup specializing in real estate.
Our specialists successfully delivered a custom, cross-platform solution with extensive, intuitive functionality, empowered through computational technology to provide the client with high-level market competitiveness.
Brief overview
Our company was contracted to handle:
- Software development
During the first stage, the team has designed a marketplace empowering intuitive property search on specific housing criteria and filters.
- ML training and implementation
During the next stages, our engineers were working towards utilizing machine learning enhancing accurate property matching and comparison.
Our solution
The project’s main purpose was designing a platform providing personalized for-rent and for-sale suggestions. With an attractive interface, intuitive navigation through pages, and filters for convenient property overview, the solution facilitates exceptional user experience.
The marketplace also encompasses analytical capabilities:
- When registering, the user can specify personal preferences going through several categories
- When registered, the user can overview personalized listings indicating accurate matching percentage
The platform is built to accelerate property search and selection by analyzing the user’s behavioral patterns. Calculating general market attributes in combination with those individual patterns, the solution can provide property recommendations to streamline business productivity.
The marketplace is configured to analyze:
- Liked and disliked properties
- Search attributes
- Personal information (age, gender, occupation, education, and more)
- Personal preferences (location, size, price, history, landscape view, the points-of-interest, and others)
AI-enabled real estate marketplace: Our contribution
As the determined requirements were changing during development, we went for the Scrum methodology. This way, we delivered business value through continuous feedback loops.
Our company successfully covered:
a. Cloud architecture
b. Application architecture (frontend, backend, data storage)
c. Data architecture
d. MVP composition
Solution architecture
At this project stage, our team took over:
1. SPA hosting
The built single-page application, based on React Native, was hosted by leveraging Amazon S3 and CloudFront to ensure constant availability and facilitate user experience.
2. Dynamic user requests, microservices
The dynamic user requests are implemented by using Amazon ECS
The microservices are managed by using Amazon Fargate container groups to ensure better scalability, network redundancy, and availability
3. User authentication
User authentication is based on the JWT standard to protect sensitive information.
4. Data storage
Data storage is implemented by using Amazon RDS PostgreSQL database to manage home data, search filters, and other relevant information.
5. Advanced search and analytics
AWS OpenSearch (Elasticsearch cluster) is implemented to store and index user activities, home data, and logs.
6. Data warehousing
ETL process is implemented to extract relevant data from available external resources and load these into Amazon RDS and Elasticsearch.
Machine learning
At this project stage, our team took over:
1. Solution design
2. Data analysis
3. Data preparation
4. Model development
5. Result evaluation
6. Automated learning and monitoring
ML stack: Pandas, Matplotlib, Seaborn, XGBoost, Jupyter Notebook, TensorFlow Recommenders
ML algorithms: Extreme Gradient Boosting, Principal Component Analysis, Singular Value Decomposition, Matrix Factorization, Clustering, Artificial Neural Networks, Natural Language Processing
AI-supported real estate marketplace: Key challenges
The challenges our team successfully resolved:
1. Ingesting data from multiple third-party sources
To resolve the challenge of different data formats and legacy source APIs, we investigated and implemented the most suitable methodologies of extraction.
2. Fitting best-fit scoring system
To handle the challenge of handling voluminous datasets, we implemented a complex multi-scoring system that considers the types and quality of datasets to automatically rank those.
Final words
We delivered a marketplace that generates personalized recommendations by utilizing artificial intelligence. Our solution, enabled by machine learning, is an on-demand product that facilitates higher competitiveness and greater customer reach, accordingly boosting business revenue.
By leveraging artificial intelligence, we delivered multiple projects:
