We found that the personal financial management systems (PFMS) in the market are not perfect. For example, Mint, one of the leaders in this field, has FOUR main problems:
- Insufficient information. Each bill, purchase, loan or transfer does not have enough details for users to know when and where the transaction happens. The status is also really vague. This lead to a bad user experience.
- Poor categorizing process. One of our member used Mint for a month. He found that lots of online merchants and payments do not belong to any categories. What Mint does is to let users update the category by users themselves. However, we believe that users would be tired of opening the app every day and update the category one by one. So users need a more efficient tool and better UX than that Mint has.
- Unable to manage categories. Though there are just several categories for users to choose and simpleness can good, users should be able to manage their own categories based on their need. We understand that the requirement of the DBMS might be extremely particular about about the details, but we hope we can make that happen.
- Bad data virtualization. The tiny chart in the Mint provides zero useful information for users to have a clear view of their finance information.
Therefore, we decided to make brand new web application for PFMS using Nessie to solve all these problems above.
What it does
Tracking, virtualizing and categorizing users' personal finance information.
- Intuitive Categorization.
- More details. Especially about merchant part.
- More choices to compare personal finance information.
- Able to predict future expenses based on all the information on the personal account.
How we built it
It is a single-page web application, running on webpack server. Nessie provides all the necessary data.
Challenges we ran into
- Purchase cannot be updated, so I had a hard time to find a way store the new categorization back to the Nessie server.
- Predict future expense is difficult, since we need to generate a reasonable model for prediction.
- New to tensorflow.
Accomplishments that we are proud of
We finished most of the parts, which are enough for us to show our ideas about UX.
What are learned
We learn about the idea of machine learning and fully appreciate and master Nessie API.
What's next for Arabica
We will implement my app to Node.JS server and run on AWS PaaS. Then, we want to build a native app (using Nativescript) and a progressive web app to increase the number of users and upgrade cross-platform UX. I want this project to be fast, responsive, reliable and powerful.