Inspiration

Every year thousands of hot dogs are toss into the garbage by Fuel Station Managers due to sanitary restrictions on how much time they are kept warm. Having a robust cook plan for every day would grant the stores a sales forecast and a money saving tool. We, as a new generation of environmental champions, seek to take advantage of the powerful tools in the data science world to help save food and protect our planet.

What it does

Our project establishes a predictive model that combines time series analysis and random forest to make reasonable predictions about hot dog sales.

Challenges we ran into

With a relatively small dataset and few variables, there is limited amount of information that we can use to build our model. In addition, relative unfamiliarity with time series models and the internal workings of R/Python packages present a real challenge.

Accomplishments that we're proud of

In less than 24 hours, We have:

Clean the original data by removing some redundant variables and adding new variables that indicate each day of the week.
Make extensive use of visualizations for understanding and inspiration.
Implement an algorithm to separate the effect of seasonality, autocorrelation, etc.
Apply a random forest model for additional analysis and future forecast.