I was at a Wegmans store earlier this afternoon to pick up some some bread and deli meat to make lunch for the week. The deli section of the store is usually busy during the early afternoons on a Sunday, and today was no different. The digital counter on the wall indicated that the servers were helping customer #42, and I was holding ticket #49. Seven customers ahead of me isn’t that long of a wait, but it was still a wait nevertheless.
While waiting for my number to be called, I walked around the deli area and noticed that the store had replaced the regular contents of one of the nearby refrigeration units. Upon closer inspection, the units now contained prepacked packages of the commonly purchased deli items (e.g., ham, turkey, bologna, cheeses). What an excellent example of efficient and responsive production planning to meet customers’ needs!
Typically, when one places an order for a particular item at the deli counter, one would state the amount of the item required (e.g., number of slices, weight in pounds) and if appropriate, the width of the slices of the deli meat or cheeses. In sum, the slicing, packaging and weighing of the items plus the printing of the price on a sticker can add up to a lot of time waiting at the deli counter for the customer. (If possible, one could place the deli order at the beginning of the grocery shopping trip and then pick up the order when one is done with the rest of the shopping to eliminate this waiting time.)
Because there is often a high demand for certain items, Wegmans have pre-sliced these particular meats and cheeses ahead of time when there are more servers available than number of customers to serve. This is efficient because there is less server idle time (i.e., higher server utilization). Furthermore, a server can now complete a service in less time because of the elimination of the slicing step, and this shorter service time also allows the server to serve more customers. This pre-slicing of meats and cheeses is also responsive because a customer’s service time is reduced (i.e., less waiting). All in all, a win for Wegmans and a win for the customer. Read More…
There were several interesting posts floating around the OR-blogosphere recently on the importance of how to communicate the results of an OR or analytics work. First up, Nathan Brixius with his interestingly titled post “Yes, Virginia, there is a difference between optimization and prescriptive analytics“:
Prescriptive analytics concerns an entire process, and this process typically involves assembling data, building models, evaluating them, and presenting the results. Optimization comes into the play for the middle two steps, but the first and last are every bit as important. In fact, many times the first and last steps – assembling the data and presenting results in a form that people understand – are the most difficult and time consuming ones.
The National Highway Traffic Safety Administration (NHTSA) recently conducted a study (“The Effects of External Motivation and Real-Time Automated Feedback on Speeding Behavior in a Naturalistic Setting“) on whether financial incentives would make drivers drive under or close to the speed limit.
In this field experiment, the authors tested an alerting system and a monetary incentive system with the objective of reducing speeding more than 5 mph faster than the posted speed limit.
The financial incentive works as follows: A driver who stayed within the speed limit for an entire week would receive $25. Each time that a driver exceeded the speed limit by 5 to 8 mph, the $25 reward would be reduced by $0.03. If a driver exceeded the speed limit by at least 9 mph, the reward is reduced by $0.06 — twice the $0.03 penalty.
The results of the study showed that the incentive system significantly reduced excessive speeding, which should also reduce speed-related crashes. In a related NPR article (“GPS Study Shows Drivers Will Slow Down, At A Cost“), Ian Reagan, one of the authors of the study and a traffic safety researcher at NHTSA, provided an interpretation of the results: Read More…
It should not be much of a surprise that large corporations have been the early adopters of analytics as these organizations are always on the lookout for ways to make themselves better (i.e., more efficient, more profitable, etc.). As such, many analytics stories have focused on these large organizations and their successes.
It was refreshing then to read a piece on analytics and SMBs (“Why Small and Medium Businesses (SMBs) Are a Big Opportunity for Business Analytics“). In the article, the author writes that SMBs can benefit just as much as their larger siblings by using analytics.
It is well known that in a successful data-driven corporation, everything starts at the management level. The management has to embrace analytics and then trickle it down throughout the entire organization. SMBs are no exception in this regard. The big advantage of SMBs is the fact that their organizational structure is more shallow and narrower in size. For this reason they are usually quicker to buy into analytics.
It is easier (in theory) for SMBs to adopt analytics because their size allows them to be nimble and to quickly adjust however necessary to something new. There is one important factor though that cannot be overlooked: the cost associated with acquiring the analytical tools available on the market. Read More…
A recent blog post on the game “Deal or No Deal” (DOND) by Barry Hughes at Game Theory Strategies got me thinking about whether the DOND endgame is essentially a version of the Monty Hall “Let’s Make a Deal” problem.
In the original U.S. version of DOND, every game begins with 26 cases each containing different dollar amounts with the largest prize being $1 million. The contestant chooses a case, which he believes contains the $1 million prize, and the game continues with the contestant opening the remaining 25 cases one at a time to reveal the dollar amounts in those cases. At various points during the game, the banker will make an offer for the contestant’s case and the contest has to decide whether to accept the banker’s offer by answering either “deal” or “no deal”. If the contestant decides to not accept any of the banker’s offers throughout the entire game, the contestant eventually reaches the endgame, where he has the choice of keeping his case (and taking home whatever prize is contained in it) or to swap it for the last remaining unopened case.
Suppose that during a game, the contestant reaches the endgame with the $1 million prize still on the board. Should the contestant swap cases? Read More…
In a previous post, I wrote that the Wizard’s problem of deciding when Dolphy Day should occur is similar to the secretary problem in operations research. The optimal strategy is to observe the first 36.8% of the candidate days for Dolphy Day and then choose the first day that is better than all of the first 36.8% of the days observed. The ideal Dolphy Day should be one that is warm and dry.
Now that the semester is over, let’s evaluate how well the Dolphy Day Wizard performed compared to the optimal strategy. I first defined the set of candidate days for Dolphy Day to be each school day from March 20, which was the first day of spring, to May 7, which was the last day of classes. During this time period, there were a total of 32 candidate days.
The charts below show the high temperature and amount of precipitation for each of these 32 days. The data (DolphyDay2012weather.xlsx) was obtained from www.wunderground.com for Syracuse’s Hancock International Airport. Read More…
Anybody living in the United States knows that the day after Thanksgiving is a day dedicated to shopping. Black Friday, as the day is called, is believed to be the day when a company’s bottom line turns black (i.e., profits) from red (losses) as consumers begin their purchases for the holiday season. Because of this, it is vital that companies have the necessary amount of inventory to meet the demand so as to not lose any potential sales due to shortages. Unfortunately, the biggest electronics retailer in the country did not get the memo this past year.
In the article “Best Buy can’t fill some online orders for Christmas,” the Minneapolis Star Tribute reports:
Best Buy gave its online customers just about the worst news possible four days before Christmas: Your order has not been filled.
The Richfield-based retailer said in a statement Wednesday afternoon that it will not be able to process some of its online orders by Friday, including some made the day after Thanksgiving.
“Due to overwhelming demand of hot product offerings on BestBuy.com during the November and December time period, we have encountered a situation that has affected redemption of some of our customers’ online orders,” it said. “We are very sorry for the inconvenience this has caused, and we have notified the affected customers.”
Didn’t somebody at Best Buy consider linking the customer ordering module to the inventory module so that when the total number of orders for a product is equal to the number of products in inventory a “Sold out” or “Out of stock” message would appear in place of an “Add to cart” message? Read More…
At Le Moyne College, spring time usually means one thing for our students: Dolphy Day!
Dolphy Day is the one day when students skip class to enjoy a beautiful spring day, and the day on which Dolphy Day falls is determined solely by the “Wizard,” whose identity is unknown except by a select group of people. To ensure a successful Dolphy Day, the Wizard has to ensure that he (or she) picks a day with decent weather (preferably warm and sunny). Over the years, the College has offered to provide entertainment and food for the day and because of this, the Wizard has to inform the administration days in advance of Dolphy Day.
The decision on when Dolphy Day happens is an optimal stopping problem, which Wikipedia describes thusly:
In mathematics, the theory of optimal stopping is concerned with the problem of choosing a time to take a particular action, in order to maximise an expected reward or minimise an expected cost.
I was introduced to optimal stopping problems in my Markov decision processes class, and it provided for one of the funniest and fondest memories of my Master’s program. Read More…
The Forbes article “The College Degrees Employers Want Most” reports that the top five degrees desired by large companies are engineering, business, accounting, computer science, and economics. This list of top five degrees is a result of a recent survey by the National Association of Colleges and Employers (NACE), which is an organization that “connects campus recruiting and career services professionals, and provides best practices, trends, research, professional development, and conferences.”
Wouldn’t it be something if colleges and universities offered a single degree that covers all five degrees? Guess what? There is, and it’s called operations research! O.R. is about solving problems (i.e., engineering) that often have business and economic implications using analytical models (computer science).
O.R. exists in engineering schools primarily in the industrial engineering programs. O.R. is also in business schools with management science, operations management, and business analytics. So, that covers business, accounting, and economics. Computer science? Every O.R. student should really learn how to code even at a basic level as it is a useful skill to have.
Finally, you know what else would make that O.R. degree better? An O.R. liberal-arts business degree!
The INFORMS Spreadsheet Productivity Research Interest Group (SPRIG) will be hosting a few sessions at the upcoming annual conference in Phoenix. One of the sessions will be a panel discussing “the past and future of spreadsheets for end-user modeling tool and for teaching modeling.”
Tom Groleau, who is organizing this session, is looking for a speaker to play the role of “spreadsheet skeptic” on the panel. Here’s the email from Tom:
I’m working on a session about the past and future of spreadsheets for end-user modeling tool and for teaching modeling. My plan is to have three or four speakers talk briefly (10 minutes or less) about different aspect of spreadsheet modeling and then move to a panel discussion, Q&A format. For one of the speakers, I’d like someone who teaches introductory OR/MS modeling without using spreadsheets as the modeling tool, or with minimal use of spreadsheets. You don’t have to be “anti” spreadsheets. I just want someone who feels that there are better tools for end users and for teaching and who is willing to come talk about it.
For an example at my own school, some of my colleagues use SPSS exclusively in introductory statistics while I use Excel for 90% of my course and have just an introduction to SPSS. Unfortunately, none of them are OR/MS folks so they wouldn’t be appropriate speakers in Phoenix.
If you are interested in taking on the role of spreadsheet skeptic, please contact Tom at firstname.lastname@example.org.