Part of the ParkHero Project
Predicts Disneyland wait times and FastPass availability by utilizing a deep neural network regression model. The code is part of this Docker Image that I run nightly to generate graphs like this:
| Data Name | Data Type | Description |
|---|---|---|
| RideID | Categorical | A unique identifier for each attraction |
| OpenHour | Numerical | The hour (local tz) that the park opened |
| HoursOpen | Numerical | # of hours the park is open for |
| MagicHours | Categorical | If the park is allowing early entry for hotel guests |
| BlockLevel | Categorical | What level of pass is required to enter the park this day |
| CrowdLevel | Numerical | The predicted crowd level of the park today (0-5) |
| Temp | Numerical | The feels like temperature in farhenheit |
| IsHoliday | Categorical | If close to a holiday, this is true |
| HoursSinceOpen | Numerical | The # of hours the park has been open today |
| DOW | Categorical | The day of the week |
| Month | Categorical | The month |
| Year | Categorical | The year |
Output will either be the FastPass or wait time of an attraction (depending on config)
First, populate config.env with the proper configurations. A Dark Sky API secret and a Google Cloud Geocoding API key must be provided. More importantly, the model will need a lot of data. The data I have in my database is in the DisneyData repo. After setting everything up, run the container using:
docker pull jester565/trainfastpasses
docker run --rm --env-file config.env jester565/trainfastpassesTo make changes to the script, simply pull this repo, make changes, and run docker build -t traindisney .
Use the command docker run --rm --env-file config.env traindisney to run the modified code.