# Category Archives: INFORMS Monthly Blog Challenge

## Can We Use Social Networks to Identify Poor Decision Making?

While suffering through the usual air travel woes recently, I felt compelled to tweet my feelings:

I was pleasantly surprised to see that Paul Rubin and Matthew Saltzman followed up with an interesting exchange:

Although I tend to agree with Paul that an outsider probably does not have enough information to decide whether or not the actions he/she sees are reasonably good given the situation (especially when it comes to the incredibly complex world or airline operations), I like Matthew’s general idea of an experiment to identify whether or not the outcome of a black-box decision making process is “good”.

Can that be done? Can we observe a black-box situation (or process) long enough to be able to tell whether the analytic machine inside the box could do better?

A large number of people carry smart phones these days, with constant connectivity to the internet. Twitter, Facebook, and FourSquare (to name a few) can determine our location, and there are other Apps that tell us which of our friends (and even non-friends) are close by. With all of this connectivity and location awareness, we can think of human beings as sophisticated sensors that collect and share information. We see a fire, a car accident, a traffic jam, an arrest, a fight, and immediately share that information with our network. In addition, human sensors are much better than electronic sensors because they can detect and interpret many other things, such as: the mood in a room (after the airline changes your gate for the third time), the meaning of an image, and so on.

Consider a hypothetical situation in which a crowded venue has to be evacuated for whatever reason. Perhaps some exits will be blocked and people will be directed to go certain places, or act a certain way. Human observers may notice a problem with the way security is handling the situation from multiple locations inside the venue, and from multiple points of view. The collection of such impressions (be they tweets, Facebook status updates, or something else) may contain clues to what’s wrong with the black-box evacuation procedure devised for that venue. For example, “avoid using the south exit because people exiting through there bump into those coming down the stairs from the second floor and everyone has to slow down quite a bit.”

In a world where Analytics and OR specialists struggle to convince companies to try new ideas, could this kind of evidence/data be used to foster collaboration? “I noticed that you did X when Y happened. It turns out that if you had done Z, you’d have achieved a better outcome, and here’s why…”

Is the airline example really too complicated to be amenable to this kind of analysis? I’m not sure. But even if it is, there may be other situations in which a social network of human sensors can collect enough information to motivate someone to open that black box and tinker with its inner workings a little bit. Those of you working in the area of social networks might be aware of something along the lines of what I (with inspiration from Matthew) have proposed above. If that’s the case, I’d love to read more about it. Please let me know in the comments.

1 Comment

## Should You Hire Security When Tenting Your House?

Last week I had my house tented because of termites. For those of you who don’t know what “tenting” is (I didn’t until about a year ago), it amounts to wrapping an entire house inside a huge tent and filling the tent with a poisonous gas that kills everything inside (and by everything I really do mean everything). Those who have been through this experience know what a hassle it is. We received a to-do list of pre-tenting tasks, which included:

• Remove or discard all food that isn’t canned or packaged in tightly-sealed, never-opened containers
• Turn off all A/C units and open one window in each room of the house
• Open all closet and cabinet doors
• Turn off all internal and external lights (including those operating on a timer)
• Prune/move all outdoor plants away from the house to have a clearance of at least 18 inches
• Soak the soil around the house (up to a foot away from the structure) on the day of the tenting
• Warn your neighbors about the tenting (so that they can keep their pets away from the house)
• etc.

We had to sleep two nights in a hotel, with two dogs, one of which had just had knee surgery. What an adventure!

The main point of concern was that the house would stay vulnerable (open windows) and unattended during the process. On top of that, one of our neighbors told us that he knew of a house that had been robbed during tenting a couple of months ago. So we started to consider hiring a security guard to sit outside the house for 48 hours. Would that be a good idea? Let’s think about this.

Our insurance’s deductible is $2500. I assume that if thieves are willing to risk their lives (wearing gas masks; oh yeah! they do that!) to enter a tented house, they’d steal more than$2500 worth of stuff. Therefore, being robbed would cost us $2500. This doesn’t take into account that one might have irreplaceable items in the house. However, most of the time those can be taken with you (unless they are too big or inconvenient to carry). In my case, I took the external hard drive to which I back up my data, and the mechanical pencil I’ve owned and used since 1991 (yes, you guessed right, the eraser at the end doesn’t exist any more). The security company we called would charge$15 per hour for an unarmed guard to be outside our house. Multiplying that by 48 hours brings the cost of hiring security to $720. Let’s say that the likelihood (a.k.a. probability) of being robbed while your house is tented without a security guard is $p_1$ (in percentage terms; for example, $p_1$ for the White House is pretty close to 0%), and when a security guard is on duty that likelihood is $p_2$. Unless $p_1 > p_2$, there’s no point in having this entire discussion, so I’ll assume that is true. Here’s a pretty neat rule of thumb that you can use: divide the cost of hiring security by your deductible to get a number $n$ between zero and one (of course, if hiring a guard costs more than your deductible, don’t do it!). Unless the presence of the guard reduces your chance of being robbed ($p_1$) by more than $n$, you should not hire security! (Later on, I’ll explain where this rule comes from.) For example, in my case 720/2500 is approximately equal to 29%. If the chance of being robbed without security is 30%, unless hiring a guard brings that chance down to 1% or less, it’s better not to do it. If the value of $p_1$ is less than or equal to 29% to begin with (I live in a reasonably safe neighborhood), the answer is also not to hire security (probabilities cannot be negative). This rule works regardless of the value of $p_1$; what matters is how great the improvement to $p_1$ is. In addition to looking at the numbers, we also took into account the following clause from the security company’s contract: …the Agency makes no warranty or guarantee, including any implied warranty of merchantability or fitness, that the service supplied will avert or prevent occurrences or the losses there from which the service is designed to detect or avert. In other words, if you hire us (the security company) and still get robbed, we have nothing to lose! So what did we do? We chose not to hire security and, fortunately, our house was not robbed. However, even though the tenting instructions say that you don’t have to wash your glasses and plates after returning home, we decided to do so anyway (as they say in Brazil: “seguro morreu de velho”). Disclaimer: The advice contained herein does not guarantee that your house will not be robbed. Use it at your own risk! Details of the Analysis So where does that rule of thumb come from? We can look at this problem from the point of view of a decision tree, as pictured below. In node 0, we make one of two decisions: hire a security guard (payoff = -$720, i.e. a cost), or not (payoff = -$0). For each of those decisions (branches), we create event nodes (1 and 2) to take into account the possibility of being robbed. At the top branch of the tree (node 2), the house will be robbed with probability $p_2$, in which case we incur an additional cost of$2500, and the house will be safe with probability $(1-p_2)$, in which case we incur no additional expense. Therefore, the expected monetary value of hiring security, which we call $EMV_2$, is to spend $720+$2500 with probability $p_2$, and to spend $720 with probability $(1-p_2)$. Hence $EMV_2 = - 3220p_2 - 720(1-p_2) = - 2500p_2 - 720$ Through a similar analysis of the bottom branch (node 1), we conclude that the expected monetary value of not hiring security, which we call $EMV_1$, is to spend$2500 with probability $p_1$ and to spend $0 with probability $(1-p_1)$. Therefore $EMV_1 = -2500p_1 - 0(1-p_1) = - 2500p_1$ Hiring security will be the best choice when it has greater expected monetary value than not hiring security, that is when $EMV_2 > EMV_1$, which yields $-2500p_2 - 720 > -2500p_1$ $\Downarrow$ $2500(p_1 - p_2) > 720$ $\Downarrow$ $p_1 - p_2 > \frac{720}{2500}$ which is the result we talked about earlier (recall that $p_1 > p_2$). How Does Analytics Fit In? The Analytics process is composed of three main phases: descriptive (what does the data tell you about what has happened?), predictive (what does the data tell you about what’s likely to happen?), and prescriptive (what should you do given what you learned from the data?). In this problem we can identify a descriptive phase in which we try to obtain probabilities $p_1$ and $p_2$. This could be accomplished by looking at police or insurance company records of robberies in your area. It’s not always possible to get a hold of those records, of course, so one might need to get a little creative in estimating those numbers. Having knowledge of the probabilities, the calculation described above could be classified as a prescriptive phase: what’s the course of action? Hire security if (cost of security)/(insurance deductible) < $p_1 - p_2$. There is no predictive phase here because our analysis does not require the knowledge of any future event (only how likely it is to occur). Operations Research can be used in some or all of these phases. Most of what I do in my research and consulting projects lies in the prescriptive phase (optimization). Recently, however, I’ve decided to broaden my horizons and learn more about the other two phases as well, starting with some self-teaching of data mining. 13 Comments ## Find Out What Happens to Mr. Lovr I’ve had a number of people tell me that they like my Valentine’s Day post, but many of them did not solve the Excel Spreadsheet. That’s the most important part of the post! You have to see what happens to Mr. Lovr! There’s no set-up necessary; just open Excel Solver (it’s an add-in you have to enable in Windows and a separate program in the Mac), and click on the “Solve” button. You’ll be glad you did. Leave a comment Filed under INFORMS Monthly Blog Challenge, Love, Valentine's Day ## Transporting Flowers with Love Mr. Lovr, a lonely gentleman, does not want to spend Valentine’s day alone in 2011. As one of his New Year’s resolutions, he intends to send roses to nine of his lady friends. Being an avid procrastinator, however, he waits until the last minute and finds out that only eight flower shops in his city still have roses available. For lack of better names, let’s call those flower shops F1, F2, F3, …, F8. The number of bouquets of roses available at each shop are, respectively, 4, 4, 3, 2, 2, 2, 2, and 1. Based on how well he knows each of his potential valentines, whom we’re going to call V1, V2, V3, …, V9, he calculates how many bouquets he needs to send to each of them to increase his chances of going on at least one date. The numbers are, respectively, 3, 2, 2, 2, 2, 2, 2, 2, and 3. At this point, a light bulb goes off in Mr. Lovr’s head, and he remembers from his Operations Research class that this is a transportation problem. But there’s something missing…Ah! The costs! He calls each of the eight flower shops and asks how much it would cost to ship one bouquet of roses to the addresses of each of his nine lady friends. He then compiles the following table of costs (in dollars): He also remembers that because the total supply is equal to the total demand, he can write all of the constraints in this problem as equalities. Essentially, he has to say that, for each flower shop, the number of bouquets that it ships has to be equal to the number of bouquets that it has. Similarly, he needs one constraint for each valentine saying that the number of bouquets that they receive has to be equal to the number that they want (according to his estimates above). To avoid suspicion, he also decides that it’s better for each flower shop to send no more than one bouquet to the same person. So far, so good, but he needs a specific shipment plan because he’s running out of time. He opens up an Excel spreadsheet and creates the following layout of cells (he chose pink to make it more romantic): Somewhere else in his spreadsheet he also typed the table of costs shown above. To his surprise, he even remembered to use the SUMPRODUCT function to calculate the total cost expression. He clicks “Solve” and finds out that the cheapest way to send all 20 bouquets will cost him$38. Not bad…but wait a second…something amazing happened! He cannot believe his eyes! The optimal solution exhibits a very curious pattern! Could it be a Valentine’s Day miracle? Could it be the power of love? If you want to see for yourself, download Mr. Lovr’s spreadsheet, open Excel Solver, and solve the model (no setup necessary, just click the “Solve” button). (NOTE: this spreadsheet has been tested with Excel 2007 under Windows XP, and with Excel 2008 under Mac OS X Snow Leopard. I’m not sure the “trick” will work in earlier Excel versions. For a free download of Excel Solver for Mac OS X, go here.)

Thanks to my love for giving me the idea for this blog post.

## The Joy of Baking (Optimally)

‘Tis the season of baking all kinds of things: cookies, cakes, breads, brownies, pies, and my favorite Brazilian dessert “pudim de leite moça“, which is depicted below. Click here for the step-by-step recipe.

Many OR bloggers, such as Laura McLay and Anna Nagurney, actually enjoy baking, and they both have written posts on the subject (e.g. here and here). I happen to include myself in this group and, yes, I made the pudim shown above (using  my mom’s recipe).

My goal today is to approach the art of baking from an optimization point of view. Let’s say you have a long list of items to bake. Perhaps you’re hosting a mega party at your house, or you’re helping your local church or favorite charity with their holiday cooking. You have an oven that can only fit so much at a time (think of area, or volume). Each item to be baked occupies some space in the oven and needs to bake for a specific amount of time. In what order should you bake your items so that you finish as soon as possible? (Side note: it may not be obvious at first sight, but this is the same problem faced by a container port that needs to decide the order in which to unload cargo ships.)

In the OR world, this is a job scheduling problem with a cumulative machine. The jobs are the tasks to be performed (items to bake), the machine (or resource) is the oven. We say the oven is cumulative, as opposed to disjunctive, because it can deal with (bake) multiple items at a time. The unknowns in this optimization problem are the start times of each job (when to begin baking each item). The objective is to minimize the makespan, which is defined as the finish time of the last job (the time at which it’s OK to turn off the oven). Finally, this is a non-preemptive problem because, typically, once you start baking something, it stays in the oven until it’s done.

This problem occurs so often in practice that the Constraint Programming (CP) community created a global constraint to represent it. It’s called the cumulative constraint (what a surprise!). Here’s a reference. For example, let’s say that we have a $10$-cubic-foot (cf) oven and we need to bake five items. The baking times (in minutes) are $20$, $25$, $40$, $30$, and $30$. The space requirements in cf are, respectively, $6$, $4$, $5$, $6$, $4$. If the time at which we begin baking item $i$ is denoted by the variable $s_i$, we can write the following in a CP model:

$\mathrm{cumulative}([s_1,s_2,s_3,s_4,s_5],[20,25,40,30,30],[6,4,5,6,4],10)$

The above constraint makes sure that the start times $s_i$ are such that the capacity of the oven is never exceeded. To minimize the makespan, we have to minimize the maximum among $s_1+6$, $s_2+4$, $s_3+5$, $s_4+6$, and $s_5+4$.

It’s easy to incorporate some real-life details into this model. For example:

• Not every item will be ready to go into the oven at time zero. After all, you’re making them as you go. To take care of this, add a ready-time $r_i$ (i.e. a lower bound) to the appropriate variable: $r_i \leq s_i$.
• If a given item does not occupy the entire oven, but you still prefer to bake it alone, just artificially increase its space requirement $c_i$ to be equal to the oven’s capacity $C$.
• If you’re baking both savory and sweet things, you probably don’t want to mix them up in the oven. In that case, simply solve the problem twice.
• If, for some reason, item $i$ must be finished before item $j$ starts baking (e.g. they need different temperatures), just include the constraint $s_i + p_i \leq s_j$, where $p_i$ is the baking time of item $i$.

We could, of course, have approached this problem from an Integer Programming point of view. In that case, we’d have binary variables $x_{it}$ that are equal to 1 if you start baking item $i$ at time $t$, and equal to zero otherwise. For more details on this formulation, including model files and some pretty tough instances, take a look at the CuSPLIB web page.

In the spirit of holiday baking, I will close with some pictures of past baking jobs ran on my house’s machine (a.k.a. oven). Enjoy! :-)

Key Lime Pie