Nutri-Cart

food love

Inspiration

As college students who are experiencing taking full responsibility for our own health, wellness, and finances for the first time, we have noticed that many of us and our classmates struggle with balancing nutrition, time, and money on top of work and our studies. The ability to pursue a healthy lifestyle is critical and defines many aspects of our happiness and wellbeing, yet we know so little and have so little time to shop for foods that are yummy and cheap.

We noticed that even our parents, who have lived in the same city for so many years, have to run between multiple different stores to scope out where to buy the cheapest and healthiest products. Bread at Pavs, Rice at Costco, Milk Chocolate Peanut Butter Cups at Trader Joe’s… but are these really the best deals, or just based on vibes? This made us realize that grocery shopping is not just difficult for us college students – it’s tough for everyone!

We wanted to build a personalized app that helps people grocery shop – whether they’re college students navigating the aisles for the first time, parents trying to raise healthy kids, or individuals shopping on a tight budget, Nutri-Cart fits seamlessly into every lifestyle.

What it does

When you first register your Nutri-Cart account (or any time your desires change), you can take our questionnaire involving a series of questions on your lifestyle, dietary needs, and budget goals. Then, every time you add items to your shopping cart, we find the best grocery stores for your preferences, whether it be maximizing nutrition, affordability, quality, or all three!

Grocery stores within a 20-mile radius will be ranked according to the price or nutritional values of the items on your shopping list. Whether you are on a sodium diet or cannot eat high cholesterol foods or if you need to replenish certain vitamin levels, the best options for you will be recommended by Nutri-Cart.

We also created a machine learning algorithm that takes in a series of questions (i.e. age, gender, frequency of physical activity, etc.) and predicts whether an individual has high risk of cardiovascular disease (95% training accuracy, 91% validation accuracy). We plan on integrating this with the website to allow an additional opt-in questionnaire to predict a series of pathologies strongly linked with diet (i.e. cardiovascular disease, Type 2 Diabetes, etc.) and provide recommendations on healthy foods and recipe ideas.

How we built it

For the front end, we used HTML, CSS, Javascript, and Bootstrap. For the back end, we used Firebase and Javascript. For our machine learning model, we used Python.

Challenges we ran into

Some challenges we faced included the initial learning curve with backend frameworks and Firebase as well as some un-familiarities with experiencing our first hackathon. We had trouble allocating how much time to work on different things and ended up needing a lot of check-ins to rescope given the time we had left. We also had multiple team members working on the same tasks in the beginning as some tasks such as data retrieval depended on the data pushed, which hindered our progress. The check-ins were helpful to keep us on track, however, and re-evaluate our goals often. None of us really knew what to expect, but this project turned out quite well and was a great learning experience. Due to the short sprints we had, we faced certain bugs that we could not easily fix and found it more efficient to restore a previous version using git. Hence, continuous integration was very helpful.

Accomplishments that we're proud of

We’re proud of how much we’ve learned - from a group who had never seriously developed a website before to friends who developed a project reflecting a serious concern in our community, it wasn’t easy but we put all of our past 36 hours into it. Moreover, we think that one reason such a grocery shopping app has not been implemented rigorously already is due to the lack of access to sufficient preprocessed data. Hence, our creative use of data in order to create shopping lists based on user preferences is something we are proud of. Our app is not hindered by the difficulty of web scrapping and access to nutritional data for grocery store items and restaurants. For this reason, we think we will sleep very well tonight. A healthy tummy is a happy one.

What we learned

Hackathons are more tiring than 14 hour flights to China. We eat a lot of snacks when we are stressed. String cheese is really good. REALLY good. The food we ate during the Hackathon definitely was not nutritious and definitely did not align with the objective of our site. Maybe we should take our own preferences survey. However, it was free! So perhaps it evens out. The power of Git!

What's next for Nutri-Cart

We want to use collaborative filtering to predict new grocery items to recommend to users (if you give a mouse a cookie, he’ll want some milk to go with it). We also want to expand this into the realm of recipe suggestions and meal planning, since we think this could be particularly useful given our app’s demographic.

We also want to use Google Maps API to gather more data on local grocery stores (specific to the user), food pantries, food delivery services, and Farmers’ Markets to add to our site. Specifically, we want this to be location-specific and work with a user’s preferred mode of transportation (i.e. accessible via public bus) and ideal travel distance.

We additionally would be interested in expanding the capabilities of our site beyond supermarkets into restaurants. We’ve noticed that it’s extremely difficult to identify whether restaurants fit with our budget and nutrition goals (for example, many of them don’t list calorie information), so it would be helpful to identify the most nutritious and cost-effective restaurants for one’s cravings.

Currently, our platform enables users to choose between prioritizing preferences such as price or nutrition through a dropdown menu. Based on their selection, we provide our recommendations as an ordered list. Looking ahead, we aim to enhance our algorithm for greater precision tailored to user preferences. For example, if a user values nutrition over price, our algorithm will factor in this weightage when generating recommendations.

Note:: None of our team members have competed in a Hackathon before, so we believe that we qualify for the Beginner Division of Hacktech.

Research: https://www.cdc.gov/chronicdisease/resources/publications/factsheets/nutrition.htm https://firebase.google.com/docs/guides https://fireship.io/lessons/the-ultimate-beginners-guide-to-firebase/ https://www.youtube.com/watch?v=PBcqGxrr9g8 https://www.w3schools.com/js/ https://getbootstrap.com/