TransacTrust

Inspiration

Seeing the rise of fraud has been alarming as it affects the common individual who might have to attend to their families to the companies that may be on a brink of a financial crisis impacting the lives of many is a dire future that we currently face. Although there are fraud solutions available, a lot of them stem from large corporations that have vast amounts of data to train their own programs. We wanted to create a project using data from the Bluehill API to create a fraud detection product that can be easily applicable to any commercial business.

What it does

Using the Bluehill API, we are able to gather data on transactions by customers in the past. From there, we take in any new transactions and using the past data we can find potential cases of fraud and even assign scores to customers based on their transaction histories. All of this allows companies to be more aware of potential risks in transactions with customers.

How we built it

We invoked BlueHill API to get historical data and we processed the data into a dataframe. We created a classifier to determine whether a transaction is suspicious or not. We created our own API so companies could use the TransactTrust as a metric to determine if their transactions are suspicious or not. We then created a website as an endpoint to output the result of the API.

Challenges we ran into

We had a hard time finding representative data from the API. We also encountered issues integrating React with Flask. More specifically, we had issues sending POST requests to our server. We also encountered problems in determining the metric if the input is fraudulent or not.

Accomplishments that we're proud of

We developed a full stack project under limited time constraints. We became more familiar with different kinds of development tools like React and Flask. Since our team was created at the very last minute, we were able to collaborate well and pulled a complete project in the end.

What we learned

We learned how to create an API that accepts requests like POST which contain JSON data. We learned how to create forms on React.

What's next for TransacTrust

There's a lot to do: the website is barebones in functionality, the algorithms we used could be refined, and system overall can be slow at times and hasn't been tested on large amounts of data. However, we got a great start this weekend, and we'll look forward to adding more soon.

Built With

• css
• flask
• javascript
• python
• react