New York City Property Pricer

Inspiration

Our group had a desire to work on a machine learning project. Two of us were new to machine learning and the other had limited knowledge and wanted to learn new machine learning techniques. All of us had interest in a project involving financial prediction. We also needed a strong dataset as to build a thorough model. We decided on creating a tool to predict New York City Property Prices.

What it does

The tool predicts real estate prices in New York City given past data. It will take in a location in New York City (broken down by borough and neighborhood), the building class of the building in question, the year it was built, the number of residential units, and the gross square feet, and predict a purchase price.

How we built it

We used traditional Python machine learning libraries such as Pandas, Numpy, Matplotlib, Seaborn, and Scikit-learn to create the model. We then used tkinter to create the GUI. The model and the GUI elements were broken down into their own classes, with a separate .py file used to integrate the two together.

Challenges I ran into

The main challenges were learning how to use new libraries as well as new tools within the libraries. We wound up basing our model on random forest, something none of us had experience with. We also learned how to use tkinter for the GUI, which none of us had done either. This culminated in a learning experience for all team members.

Accomplishments that I'm proud of

The greatest accomplishment was learning how to used the tools and libraries well enough to create a working application in less then a week (after planning was completed), while we are so early in this program. We used a language which isn't used with the first core courses in the program, but used concepts that were taught to complete our code.

What's next for New York City Property Pricer

To expand upon the concept and potentially make this application look and run a bit more professional. We plan to improve it by asking user to give us the sale price if they make a successful sale, and then dynamically improve the training data for the ML model.

Built With

• python
• tkinter