New York City started providing free breakfast and lunch to all of its 1.1 million K-12 students for the 2017 school year. In this talk, data scientists at the Mayor’s Office of Data Analytics will show how they worked with the Department of Education to solve a combinatorial optimization problem to save the City millions of dollars while integrating every school into the free lunch program.
The NYC Department of Education had to optimally organize the thousands of breakfast and lunch programs in its schools in order to provide universal free breakfast and lunch service. Doing it by brute force would take far longer than any school year, but the Mayor’s Office of Data Analytics built several Monte-Carlo method optimization algorithms to get there much more quickly. We will walk you through the problem, the analysis techniques we used (stochastic hill climbing and simulated annealing), and demonstrate how Pandas and Python's parallel processing capabilities turned an intractable problem into one solvable on a laptop. This talk is intended for anyone with an interest in how the government uses data science, anyone with an interest in optimization algorithms, and even people relatively new to the Python data ecosystem looking to implement an analysis from start-to-finish.