Last week was the 4th edition of the programming contest Equinox, a hack day and hack night for the CDO unit (Chema Alonso’s unit) at Telefónica.
For the competition you can present anything you want as long as you made it work. So we made a data science team to design an intelligence elevator system in R.
Main objective was to optimize in real time the waiting times of the elevator’s queue. The structure was created based on three main modules: a simulation system, an optimization system, and a visualization system.
The problem was initialized for a building with 9 floors, 6 elevators and 150 employees per plant approximately. Average transit times, elevator mobility between plant and plant and opening / closing of doors times were estimated based on a real case and a Normal distribution with a mean of 9:00 is assumed for arrival times. Initial datasets were generated randomly to train the optimization system. Algorithm chooses the best elevator in real time for minimize waiting times efficiently. And finally, an app had been developed to visualize the results of the previous modules, showing route of each elevator, assigned elevator for each arrival, arrival times distribution and final data table with all movements. User can regulate a slider to simulate elevator movement in real time.
We used dplyr, shiny, and ggplot2 libraries. You can find the code here: https://github.com/PaulaLC/equinox