--- Elevators!!!---
Okay so I hate the elevators in our building (they are old slow, terrible, and always get “stuck” at the top floors) and that got me thinking about ways to make them better. What if I could forecast the floor that an elevator should be on? So in this project I am going to try and take some (fake made up by me) elevator data and try to forecast what floor it should stop on based on the time of day.
########################################################################################### And this is an excuse to learn some R (it will be obvious soon that this is my first project in R).
My end goal is to be able to take one days worth of data (when people get to the elevator, what floor they get on) collapse that (by mean) to 144 10 min "sections" of time, round those answers which should lead to me having this nice data set. Then run a seasonal ARIMA model to predict where the elevator should be on the next day.
I will try to be as step by step in my actual script as possible, but pretty much my steps are.
- Proof of concept with some super random data
- Import some new data and adding some columns of variables (some cleaning)
- Collapsing my data into 10 min averages and further cleaning
- Running an ARIMA model on this new data
- Showing the number of floors saved
---Next Steps---
- Getting actual people getting on elevators
- Trying out different kinds of forecasting models