This directory provides an implementation of the methods for rank-based dynamic assortment optimization models.
Julia 0.5
Distributions 0.13.0
NPZ 0.2.0
Instance is the user-instantiated type, describing the parameters of each generated instance of the dynamic assortment problem: number of products, customer types, distribution of the number of arrivals, and choice model.
instancegen.jl: methods relative to the instance generation, logs, and revenue evaluation.
heuristics.jl: implementation of heuristics and approximations including discrete-greedy, gradient-descent, and local search.
approx.jl: implementation of heuristics and approximations for the general rank-based model and nested choice model.
Usage examples are provided in main.jl. A benchmark of instances is provided in the folder instances and the corresponding method outputs are provided in the folder outputs.