Foodie

 The Question: 

As modern consumers, we benefit greatly from recommendation applications, but how can we accurately recommend what restaurant a consumer can eat at based on their budget and food preference?

Motivation and Background:

A common problem that all Americans face in today's society is finding the right place to eat based on the budget they are working with. Our food platform, FOODIE, strives to solve this problem and go one step further by not only finding the consumer with the right restaurant based on their budget but also tailoring that decision based on what the customer is in the mood for. We live in an age where computers help us make decisions, with applications such as Netflix recommending movies or television shows, and Spotify recommending music genres and artists based on your usage. By recommending users where to eat based on their budget, our users can avoid the issue of spending more time talking about where to eat rather than actually getting food.

While looking at the market for restaurant recommendations as of right now we can see that they are merely restaurant suggestions but not specific recommendations. We can see applications such as Open Table, Trip Advisor, and Google give you restaurants near you and what type of food that you are looking for but don’t let you know how much it's going to cost. Foodie gives the user the ability to decide what they are eating based on what type of food they want, what type of restaurant they want to go to, and what their budget is. We believe that finding the right place to eat should be as easy and 1,2,3 (Food, Restaurant, Budget). There shouldn’t be any reason why people believe that finding the right restaurant is hard, whether it's going out for a burrito or if it is deciding on where to take that cute girl on a first date.

Ethical and Social Implications: 

Although we don’t plan on publicly launching our product, if hypothetically we did, we might see a change in the restaurant economy in Boulder. Since Boulder is mostly made of college students, we feel our product would give attention to restaurants that are usually cheaper than others since typically college students have less income than those who are not. Restaurants that have been plagued by biased reviews might see more commerce since the user won’t be able to make a judgment off the reviews. While some businesses would boom because of it, others might start falling behind because of a stigma against them for being overpriced or not worth what they are charging. Normally, restaurants would see a steady decrease in business if their food is not worth it, but our platform could speed up the process. Although we are still figuring out feature details, one aspect we talked about was the ability for users to include their food allergies. If we have this feature, we will be able to give an opportunity to those who decided against a restaurant because they are unsure if they contain the allergy. This feature would then be more inclusive to a group of people that are often not thought of when creating a restaurant finder platform.

Related Work: 

While searching for other systems that focus on recommending we found that there are other restaurant recommendation algorithms that focus on the specific type of food that the restaurant does. This algorithm could be repurposed to focus on the specific amount of money that the user wants to spend, then once the budget is decided the user will be able to see what type of food they want. While reading through “How to build a Restaurant Recommendation Engine (Part-1)” by Nagesh Singh Chauhan he states that there are 3 different types of recommendations, Simple, Content-based, and Collaborative. The specific type of recommender that would be useful for the type of project that we are working on are both Content-based to see the type of food they like and what else they would like based on what they have reviewed before, and then Collaborative filtering engines would use the metadata to see what they have liked before to see what they will like in the future. For the scope of this project, we will be using something similar to Content filtering. Since we don’t have a user login system, it would be hard to make predictions based on previous orders, however, we can make recommendations based on previous user’s experiences using other platforms such as Yelp. One aspect that we plan on improving on is the budgeting of the user. Currently, Yelp and Google give you an estimate typically with a scale from one dollar sign all the way up to four. While this method gives a decent estimate of how much a user spends, we want to focus more on the issue by giving specific meals to the user, so they more realistically plan how much they’re gonna spend.

Method: 

In order to get to the prototype of our website, we will first need local data from Boulder. Although there are tons of restaurants in the area,  we plan on web scraping ten to fifteen of them to make sure that we collect all of the meals with the proper attributes such as price, reviews, and possibly nutrition. The data would then be loaded into a server that allows our program to access the information. Once we have our data collected, we plan on creating a web application that would take the user entered attributes and return restaurants that fit their needs. While we are still working out on what attributes help the user come closer to a decision, our focus will be more on the individual meals of the restaurant instead of the restaurant as a whole. Through personal experience, restaurant reviews tend to be broad, whether talking about the actual location, parking, service, etc. instead of the actual food. We will do some preliminary research on other sites that recommend restaurants to see what kind of design implementation they have and see if there is room for improvement. After completing these three parts, we should be able to combine them for a fully interactive web application. Although there isn’t an exact evaluation method, if our results are different then our competition and the test users are satisfied, then we will know if our application is good. One of our limitations for this project will be the data we collect. As we stated earlier, we plan on collecting around the Boulder area, so there is an implicit bias towards the restaurant selection in our area. Also, while we all have some programming experience, none of us have the experience of building a script that works in a browser application so we will have tons of testing, but it somehow doesn’t work, then we are limited in our final product. If all the parts don’t come together, we can at least do a ‘wizard of oz’ prototype where we have the site completely designed, but have the user enter in their query into a separate script that would give them their meal.

Deliverable: 

For our final deliverable we create a website that allows users to interact with key features of the application for new users such as setting favorites and restrictions. As well as key features for existing users such as saved favorites and restrictions and the process of selecting a restaurant for that meal. Our website while not fine tuned with design allowed us to get a feel for if our application was viable and would it feel different than Yelp and other restaurant recommender applications. The website consisted of a login page, new-user flow, and existing-user flow.

The new-user flow consisted of setting up your favorite types of food, and restrictions. We wanted our app to focus on these things because we felt other apps were lacking when it came to recommending food we liked, and could actually eat.

The existing-user flow consisted of skipping selecting the food, and restrictions because we already have it saved. Next, users select their budget for the meal. Then their comfort settings. And then using all the parameters the user has given us, recommend 2 restaurants that the user enjoys. The user can then choose to skip over them or not. 

We were only able to test our functionality with a few people since the way our app was built. But from the tests we were able to find that users did think that this was different than other apps, and would accomplish something new and helpful.

Conclusion: 

Our findings from the application is that the people that have looked at our application really like the concept, and they believe that it would be extremely useful in their everyday life. Foodie allows its consumers to have their preferences and budget come first when deciding where to eat but takes away the confusion of choosing where to eat. We believe that if we are able to put the code that we have written with the UX design that we have mocked up through Bubble.io we could have a great working website that could be eventually changed into an application. With all the positive feedback and constructive feedback that we have received over the course of the semester we are able to take that and finish up our first working prototype that we could release to the general public. 

Throughout, the semester the three of us all had our areas of emphasis and we believe that has helped us create Foodie, Malik was the Software Developer and Chief Technological Officer, Steven was the Graphics Designer as well as the Chief Executive Officer, and finally Spencer was in charge of Business Development and the Chief Marketing Officer. By having these positions in the project group it helped us stay organized on what tasks that we needed to get done and allowed us all to become experts on the area that we were working upon. 

If we could give some advice to the students that are taking this class next year we would say to make sure you stay ahead of deliverables and work collaboratively as a team on something all of you are passionate about. By staying ahead of deliverables you are making sure that you for one have a great project, and for two it allows you to stay ready for any bumps in the road that may pop up. Our group was faced with one of the biggest challenges of our college experience when classes got switched to online meaning that if we were not ahead of the curve we may have struggled to pull our project together. By having your own area of expertise in the project you are able to cover all bases when completing deliverables making it easier for you to get it done and be able to answer the answers correctly. Finally, we want to say that we recommend working hard and most importantly have fun since this is your last couple of months of the best 4 years of your life.