Statement of Work Shin Woong Sung 1. Executive Summary This Statement of Work (SOW) suggests a plan and a solution approach to find out the best mix of machines for each casino site of Lucky Duck Entertainment, which maximizes the utilization of floor space. Through data analysis and several interviews, we categorize the machines and forecast the daily profitability of each machine in each section of each casino. Using real-time transactional data and other operational data (floor space, management preference, procuring cost, etc.), the input data of optimization is generated in the form of Microsoft Access tables and stored in a company database. An integer linear programming is formulated to maximize the floor space utilization under the limited floor space and budget. We assume that maximizing the utilization means maximizing the profitability per unit area. With this optimization, we can find out the best mix of machines in each section of each casino. We will use SAS/OR, a powerful operations research software, from reading the input data to implementing and solving the optimization model. As a follow-up project, we can further design the layout of each casino site that maximizes the utilization, by using the best mix of machines from the optimization and the demographics data of loyalty members. 2. System Environment A. Machines Supported: Intel or Intel compatible Pentium 4 class processor with SSE2 instructions (minimum required) B. Memory Requirements: 1 GB minimum (more memory is recommended for improved performance)
3. Project Description Information Flow First of all, real-time transactional data and other operational data (floor space, management preference, procuring cost, etc.) stored in a company database are used to generate the input data. The database system (Oracle, SAP, etc.) exports those data as Microsoft Access tables, and Visual Basic programming for machine categorization and profitability forecasting generates the input data of optimization. Then, the input data (MS Access table) is imported to SAS datasets and the integer linear programming model implemented in SAS/OR finds the best mix of machines that maximize the floor space utilization. The result from this optimization is stored into a SAS dataset again and finally exported to a Microsoft Access table. Analytical Components A. Machine Categorization & Profitability Forecasting: Through the analysis of sample data, we found out that the daily revenue and plays are different by machine name, casino, and section, and we categorize the machine by machine name, casino, and section. For the data instance whose Plays is greater than 100,000, we assume that the data is enough to be accurate and we can use the revenue forecasted by historical distribution. For the others, we assume that it lacks the data and decide to use the vendor reported revenue. We can find average daily revenue per unit machine by dividing the forecasted gross revenue into the number of machine and the machine use time (=base date- Month ). Similarly, we can find average daily plays per unit
machine by dividing the plays into the number of machine and the machine use time. Finally, we can calculate expected daily profitability per unit machine by multiplying the two - daily revenue and daily plays. It will be used as the profitability of machine and as the parameter of the optimization model. This process can be programmed into Visual Basic code and automated. B. Optimization: For floor mix optimization, we developed the integer linear programming that maximizes the profitability. The parameters, decision variables, and the mathematical model are below.
Input Data Structure Input data is formed as Microsoft Access table and consists of four tables machine information, current configuration, management information, and forecasted profitability information. In the machine information, there are procuring cost, decommissioning revenue, required floor space, and warehouse inventory level for machine j. In the current configuration data, the number of each machine j at casino i in current machine mix is stored. In the management information, the amount of floor space at casino i, the allowable deviations from the current configuration, and the management preferences for number of machine j at each casino i. In the forecasted profitability information, there is forecasted profitability of machine j in section k at casino i as a table Output Data Structure Output data contains the results of optimization and is formed as Microsoft Access table. It consists of the optimized machine mix at each casino. 4. Analytical Components In order to perform the project work, two software products are necessary. A. Microsoft Access : for machine categorization and profitability forecasting B. SAS and SAS/OR : for floor mix optimization
5. Assumptions The more plays means the more data Maximizing the utilization of floor space means maximizing the profitability per unit area There is a budget limit (flexible) All machines require different floor space The floor space of sections in each casino is limited. The effect of the number of machines put together is ignored The profitability of the same machine name can be different by casino and by section From data analysis, we find out that the machines located in the interior or restaurant plaza section are more profitable than those located in the entrance or boundary. So, we take the two sections (profitable/less profitable) into account in the decision variables. The number of decision variable is 6480 (8 casinos, 405 machines, 2 sections) For the machines with no data, we regard the profitability of those machines same as the existing profitability on the same machine in other casino. If the section is different, we can use the formula below. Profitability of a machine in the boundary or entrance = (Profitability of the identical machine in the restaurant plaza or interior)*0.3 There is no transfer cost from warehouse to casino