Xxxxx xxxxx
Xxxxx xxxxx
Xxxxx xxxxx
Xxxxx xxxxx
Input Modeling Report
Brief Description:
The system being analyzed is the daily operations of Co-Workerate, a co-working office
space in California. When customers arrive, they purchase time in advance for how long they
anticipate to use the facility for the day. Customers can choose either open seating or book a
conference room. There are 4 different seat types for open seating and two different sizes of
conference rooms. Parties of different sizes are observed to have certain seating preferences.
Four weeks of customer data was collected to use for input modeling.
Inputs:
The pertinent inputs consist of arrival rates, party size, and time purchased.
Decision Variables:
Decision variables are variables that Co-workerate can change to maintain or improve
profit while continuing to deliver a high-quality customer experience. From the our research
questions, we can identify several decision variables: the number of seats of each seat type,
seating assignments, daily operating hours, cafe space, and internet speed.
Observable quantities:
Parties of different sizes have different seating preferences and assigning seats or
adding seats of a certain type may improve utilization. Early mornings on Monday and Tuesday
have very low volume, which may not offset the operating costs during those hours. Many
customers leave Co-workerate during lunch so additional cafe space may improve length of
stay. Lastly, slow internet speed is a primary customer complaint during peak hours, so
changing the internet package - at an additional cost - may improve the quality of the customer
experience.
Outputs and potential performance measures:
With our research questions in mind, there are a few metrics we want to track.
Performance measures include daily profit, average length of stay, and total customers served.
Seating utilization, the count of groups of each size, as well as length of stay will be observed as
state variables.
Summary statistics:
The data provided includes arrival date, arrival time, number in party, amount of time
purchased, and space reserved. The number of customers arriving in 15 minute intervals
appears bimodally distributed, with one distribution for customers arriving before lunch, and a
different distribution for customers arriving after lunch. The amount of time purchased is discrete
in half hour intervals, but the count of customers purchasing a certain amount of time is roughly
poisson. The domain of a poisson distribution must be integer values, so an integer distribution
Xxxxx xxxxx
Xxxxx xxxxx
Xxxxx xxxxx
Xxxxx xxxxx
can be created by doubling the time purchased and counting the number of customers at each
integer value. The sample mean of the half hour intervals is 3.0 and the sample mean of the
integer distribution is 5.97. The poisson
parameter is estimated by the sample mean.
Xxxxx xxxxx
Xxxxx xxxxx
Xxxxx xxxxx
Xxxxx xxxxx
Stochastic Input Models:
Number in party is a discrete distribution limited by the seating types at Co-workerate
and ranging from 1 - 8 people. The probabilities for each group size were estimated using the
sample data.
Number in Party Count Estimated Probability
(Total = 1251)
1 643 0.5140
Xxxxx xxxxx
Xxxxx xxxxx
Xxxxx xxxxx
Xxxxx xxxxx
2 172 0.1375
3 73 0.0584
4 275 0.2158
5 40 0.0320
6 20 0.0160
7 19 0.0152
8 9 0.0072
Research Questions:
A properly built model should provide insight on three major questions. First, how would
leasing additional space affect profit? Secondly, how would extending the cafe menu affect the
average length of visit and ultimately profit. Lastly, how would assigning open seating affect
seating utilization? It is important to realize none of these options are mutually exclusive. In
other words, any number of system changes can be implemented. To help answer such
questions, the performance measures of daily profit, average length of stay, and total customers
served will be measured within the simulation. According to these, decision variables like
seating and cafe space utilization may be configured.