Mosquito Repellent Sales Simulation

Simulation Modeling course project

Overview

This project is a simulation, built in AnyLogic, of the point of sales of mosquito repellent products based off of the population of mosquitoes and any mosquito warnings and advisories. This project uses data provided by the Health Regions of Saskatchewan, and 51 pharmacies located in Saskatchewan. In addition to the mosquito, point of sales, and advisory data, this project also attempts to take into account the effect statutory holidays and discounts of products, which is made possible due to the data provided. Because of the way in which the mosquito population moves in Saskatchewan, this simulation is broken into regions according to the ecozones, which are defined by the Health Region. This project uses an Agent Based Model to simulate the point of sales. Three populations of agents are used, one to model each of the ecozones for which the data was available. The agents choose whether or not to protect themselves by apply repellent, and on the occasion when they run out of repellent or they decide to buy repellent for future use, they buy more. These purchases were then tallied in the main and graphed against the real point of sales data.

Results

It was found that the advisories and number of mosquitoes seemed to have the greatest correlation to sales. Discounts and holidays has the smallest correlation with sales; although, at the beginning of each mosquito season (in middle of May) sales appeared to be mostly driven by discounts and holidays. There was an overall trend throughout all regions and years of highest sales in the end of May and beginning of June with a steady decline until the end of the mosquito season (in September). Discounts had the least effect on sales at the end of the season. This trend of declining sales throughout the season was something that differentiated the simulated POS from the real POS.