Inspiration
People spend a lot of time trying to find a home. It is one life's biggest investments and because of that, a home is a match for consumers. So the goal of the project was to create an application that provided and utilized machine learning API to learn what the consumers were looking for and help assist consumers in finding homes that consumers wanted. It is also considerable that Machine learning is not perfect but even then, it can help gather statistical data to help the realtor agent help the consumer find a home that fit more of what they were interested in.
What it does
The application uses machine learning and a neural network to take some very little information and use it to learn what the consumer wants by having the user select Like or Dislike on presented houses. As they click more and more, the aim is for the machine to hone in on a house that the user wants to buy. The aim is to possibly even have it extended further to provide valuable information to realtor agents to assist them in making decisions to help the consumer find a house that is right for them.
How we built it
We kept it very basic since we aimed more for functionality and the learning experience. The main things used is Java for the main language, Weka for some initial analysis and information, JSON for possible integration with other platforms and systems, and CSS to dress the application up.
Challenges we ran into
Time was a restriction and made some thing difficult or impossible to accomplish. Things like training a model and having access to a large data set made training and configuring the initial neuron network difficult. Also, since it is a short time, it's accuracy could be improved with better methods or algorithms that take more time to implement and train. Other things we tried such as Android studio or connecting to a Javascript interface didn't quick work out and forced us to drop some things in order to provide a functional solution.
Accomplishments that we're proud of
To be able to code a working Neural Network from the ground up and learn based on minimal input of data is really a challenge but also rewarding. To be able to create an application that can predict relatively well is something is great experience.
What we learned
Machine Learning isn't easy to learn and understand. It isn't hard to actually implement. The major part of it that makes it so difficult is the design and choices because that ultimately affects the output. Also, when restricted for time, priorities in what you want to implement changes so you must adapt.
What's next for Machine Learning Assisting Consumers
Sleep.
Log in or sign up for Devpost to join the conversation.