This talk will demonstrate how Julians can easily transform their AI output to a modern website, as well as how Julia can become the go-to language for complex web applications. DanceJL can integrate with any frontend web framework of choice, by allowing the framework to easily plugin into the AI/backend logic.
We would like to share with fellow Julia enthusiasts:
- Dance.jl: A new framework for building interactive web services.
- Propertyanalyser: An actual application of Dance.jl as part of an interactive tool for residential real estate valuation through various statistical methods.
1 - Dance.jl
We would like emphasize that this is not a competition to other packages such as Genie.jl or Mux.jl. Simply our approach is from a more broad understanding of web development, allowing traction of Julia from non-data scientists.
Features include:
- Render Dicts & DataFrames into JSON API endpoints and integrate with any frontend framework
- Simplified dymanic url rendering
- Webpack4 integration via plugin (headache for many web developers)
- Database (ClickHouse) plugin
- Tasks queue management
- Docker image to easily run project as isolated web server
2 - Propertyanalyser
Demonstration will illustrate following insights, using the Iberian urban real estate market (Madrid, Barcelona, Valencia, Lisbon) as an example:
- Geographic drill-down into property price levels through geographic heatmaps both for purchases and rentals, as well as rental yield
- Residential real estate valuation techniques (e.g. comparison of R2 from linear regression vs neural network)
- Breakdown of value drivers (e.g. pure location vs individual property features such as size, floor level, year of construction, expected rental yield)
- Identification of “hot deals” that appear undervalued compared to the local market
We look forward to exchanging ideas with fellow Julia professionals and to meeting you in person.