Please can anybody help me with my final year project?
I'm working on an assortment optimization problem in retail (Food industry), and i have a big transactional data . i have to chose only 4000 products from a set of 15000 product in a way to maximize the total revenue. I'm supposed to use an operational research approach , and the only thing that cross my mind is to use the MILP(mixed integer linear programming).
I dont have any constraint about the shelf space or the stock restriction, can i use a knapsack approach in this case ?
also i want to know if I can use multi label classification to determine the probability that a client might buy a product from a specified category?
thank you in advance for your answers.
The knapsack problem has one important restricition - the size of the packed items cannot exceed the knapsack storage capacity so it seems not to be your case. You have a restriction on the number of various products(items), not their size (weight/volume etc.). If you assiociate a xi binary variable with each product avaialble, the restriction would be like x1+x2+...+x14999+x15000
First of all, i want to thank you for answering. Yes it's just the selling price, maybe it will be multiplied by the purchasing rate of each product.
Anyway, there must be something what makes the problem difficult. I've realized that if there is just a constant revenue related parameter associated with each product, then you should just choose 4000 products (out of 15000) with the largest revenue parameters and that's all. What am I missing?
I studied order acceptance and scheduling on identical parallel machines in which the objective is the maximization of total revenue. You can find the study in my profile. I used the MILP(mixed integer linear programming which is not enough to solve the big-size mathematical programming so I used Genetic Algorithm. Big-size mathematical optimization problems usually need this kind of metaheuristic way.
Some thoughts I have: Knowing the breadth of the products, time period of the data you have, do you have store locations (lat/log's), demographic data, as well the product categories. Do we know sub-categories, brands, if the items are substitutes or complements, as well as the what the consumer decision tree (CDT)? What is the consumer buyer behavior? Do we know product space, or fixture space? MILP is a common method. Are there reasonable constraints that should be applied to make the recommendation more actionable? Interesting topic, I'm interested in hearing your progress.
- Badges
- Science topic