Assortment-Eskandari
Description
we generate a series of random test cases to illustrate the applicability of the proposed modified augmented ε-constraint method. As this is an NP-hard problem, search space will expand by increasing the size of instance data, and the computational time will increase exponentially. For this reason, we consider instances of problems where the execution time of the proposed method does not take more than ten minutes. The test cases are generated following the presented procedure. The number of products is considered to be n∈{30,40,50} and the number of customer types is considered to be m∈{100,150,200}. We sample the revenue of each product independently from a uniform distribution over [1,100]. We divided the products into three categories as follows: 1) Inexpensive products with a price range between 1 and 40, 2) Medium price products with a price range between 41 and 80 , and 3) Expensive products with a price range between 81 and 100. We have chosen a collection of m customer’s preference lists with a maximum length of 10. It should be mentioned that each preference list represents a customer type. Customer types are also divided into three categories. The first category (G_1) is customers whose preference list includes only inexpensive products. The second category (G_2) is customers whose preference list consists of only medium-priced products. The third category (G_3) is customers whose preference list includes only expensive products. We consider 30%, 50%, and 20% of customers as the first, the second, and the third category. It should be noted that if the number of customers in the first category is m_1, the number of customers in the second category is m_2, and the number of customers in the third category is m_3. Finally, we have generated arrival probabilities for each customer type by generating m_1 numbers (a_1^',a_2^',…,a_(m_1)^' ) independently from a Poisson distribution with λ=30, m_2 numbers (b_1^',b_2^',…,b_(m_2)^' ) independently from a Poisson distribution with λ=50, and m_3 numbers (c_1^',c_2^',…,c_(m_3)^' ) independently from a Poisson distribution with λ=20.
Files
Steps to reproduce
Read pickle files with python