Employee Preference Input
Heatmap of employee preferences
Shift workers prefer to work certain days over others, but can't express this with their timetable. Managers hate having to manually adjust schedules to everyone's preferences each week.
prefer.io is a gamified system where each employee receives credits to allocate to shifts depending on how much they want that shift. Placing a lot of credits on one shift is a tradeoff as you have less to allocate to other shifts you may want.
Managers can press a single button and our algorithm calculates the optimal roster. Managers can also see a heatmap of popular and unpopular times, allowing them to hire new employees that can work the unpopular shifts.
A common economics problem is the gap between a scarce resource and the demand for it. By giving employees fake credits, we essentially created a "free market" solution to solve the scarcity of shifts. The employee that want the shift the most will allocate the most credits.
How we built it
In order to solve the preference allocation problem, we created a matrix of preferences, where each row was a worker in need of a shift, and each column was a shift. We converted this into a square matrix by duplicating rows when a worker would like to work more than one shift in order to use previously solved problems in graph theory.
i.e. Row i, column j of this matrix tells you the amount of credits worker i allocated for shift j.
This essentially boils down to a maximum-weight minimal edge cover problem, which can be obtained by adapting the Hungarian algorithm developed in 1995.
Accomplishments that we're proud of
- Creating and mapping a general purpose algorithm to allocate finite resources (shifts in this case) to some amount of people. This can be used in bidding for scarce resources or other timetabling applications.
- Creating a elegant and fast web app for employees and management alike.
- Incorporating Google API sign-ups to save user data.
- Using Firebase API for the database.
- Visually representing the results with a heat-map and a recommended timetable.