• 

  Linear Regression model that takes in board_disposition and complainant_ethnicity as inputs and predicts fado_type as output.

• 

• 

Inspiration

We were inspired to make this project because we wanted a better community. In recent years, concerns of police brutality and complaints going unheard have been on the rise. In the U.S, only “Eight percent of force complaints were sustained.” What if there was a way to increase accountability? We wanted to make sure we could find out how efficient and helpful police and procedure are.

What it does

Our website takes in New York City Police Complaints data from 1985 to 2022, and it predicts the quantity of crime and the amount of time it takes to process a complaint. With this data, we can determine how much to allocate to police training so that these incidents don't happen again!

How we built it

First, we used machine learning models in python and jupyter notebooks. After obtaining a linear regression line, we made a website using Velo by Wix to create a readable report consisting of all of the most common complaints.

We will also be able to predict future trends in advance.

Challenges we ran into

There were two main challenges we ran into.

1. Linear Regression Models use numerical values as inputs and outputs. Therefore, we had to turn text data into numerical values. This led to a low R^2 value which indicates low correlation when we forcefully turn text inputs into numerical inputs. Moving forward, we hope to create a better predictive model. This can be done by either using better word embedding technology or other machine learning algorithms instead of linear regression.
2. Despite producing the graphs in Jupyter Notebook, we weren’t able to use HTML to put the graphs onto Velo. In the future, we intend to fix this issue so people can ask for specific years and get predictions

Accomplishments that we're proud of

We were able to learn more about the arduous but interesting process of building Artificial Intelligence. On top of the models, we created a website that is open to the public!

What we learned

We learned about machine learning, testing, data sets, training, and prediction models as well as graphing. We also learn a lot about the topic during the research phase.

What's next for Police Education

We hope to create a more cohesive frontend connection with actual jupyter notebook processes enlisted.

Built With

• jupyter-notebooks
• machine-learning
• python
• velo
• visual-studio