Cheap Eats | Devpost

Inspiration

Budgeting is hard. All of us have tried to start budgeting and lost track sometimes due to an overwhelming number of things to track. In this age of unlimited expenses, it is extremely easy to lose track of how to budget while tracking everything. We decided to help people ease into budgeting by targeting food: the second biggest expense after rent. Many people tend to exceed their food budget, and this is especially true for college students, who spend way more on food than they would realistically like to. This can be attributed to the psychological term ‘Present Bias’ where you tend to weigh choices with immediate rewards much more than meaningful choices. Although it is really hard to change unconscious habits, we have put our best effort into developing Cheap Eats to help manage your budget through constructive tracking and suggesting algorithms that even account for “cheat” days.

What it does

Enter a budget that you would like to spend under and explore all the countless options of individual food items near you. Track your budget as you buy food. Eating out can get expensive, but you also want to try new things while staying under your budget. Never go out to eat and spend more than what you need to again.

How we built it

We wrote automated scripts in python using selenium and other web drivers in order to scrape information about the food items. We used Google Firebase in order to serve our database and images. The Frontend was developed using Flutter. We used two open-source food datasets on Kaggle in order to train two separate recurrent neural network classifiers that tag our food as fast food and vegetarian respectively using google colab. We also wrote an algorithm that uses all your logged spending data in order to suggest an allowance on a given day. This feature helps you maximize the days you can eat out while staying under your monthly budget.

Challenges we ran into

Unable to host network image URLs properly. Getting meaningful results from the RNN Classifier and also building the suggested budget algorithm. All of it took some perseverance and couple red bull to solve :)

Accomplishments that we're proud of

Scraping the Websites and building generalized components to scrape multiple ones. Building an RNN classifier to tag our items and tuning them with the right hyperparameter. Learning all the flutter framework. Getting over the pressure of the first hackathon.

What we learned

Budgeting should be targeted one facet at a time, starting with food. This will incentivize people to begin budgeting and keep track of their finances.

What's next for Cheap Eats

Future plans we have include in-app purchasing, introducing more filters for dietary restrictions, tracking calories, and also building a constructive meal plan based on your budget and habits.