 Hello global supply trainers and welcome to the third life event of SC0x supply chain analytics. We're very happy to be here today with you. I'm in Maborrella. I'm a postdoctoral associate here at MIT Center for Transportation and Logistics. And I'm the course lead of SC0x. And today I'm with Chris Kaplis. Chris is the executive director of the MIT Center for Transportation and Logistics. And he has been your instructor during the whole course of SC0x. All right. Thank you. This is, is this the third course you've run now? Second, second. All right. It seems like three, right? Seems like a lot. All right. So what we're going to do today is a couple things. First, I'm going to give a recap of the course just really quickly to put things in context. And then also describe how that fits into the whole curriculum for the MicroMasters. Because hopefully you all will continue with this program. And then we'll talk about two problems from practice that we received from you all. I'm going to clean them up a little bit and make them easier to digest in a short period of time. We'll have a breakout. You guys discuss it. Don't come back and Inma and I will discuss it and see what we come up with. And then at the very end, we'll wrap up and talk about the next follow-on course, SC1x. So let me talk about this course. SC0x is probably the first course that most people are taking. And to be honest, it's probably the hardest course. So if you've gotten through it, congratulations. This is where we put all of the math. It's not the first course we created. We actually created SC1x first, then SC2x. But then when we needed to expand to five, we kind of followed the format we do at MIT. And we pushed all the analytics into one course. That's great because you have it all there. And so now you're equipped for everything you're going to see in the next four courses. But the downside is we hit you very hard, very fast with all this math. So the rest of the courses just apply it. Trust me, no new math will be taught in any of the other four courses. You've seen everything. Now we're just going to apply it. So what did we do? We started off with the basic idea of a mathematical function. And for some of you, this is old hat. For many of you, you've never seen this before, but you've been using it. So that whole idea that y is a function of x. And we talked about different functional forms. This is something that needs to get ingrained in you because any kind of mathematical model uses functions. You just have to get used to it. But then we really, excuse me, you can divide the course into two. The first half is deterministic methods and the second half is more stochastic or dealing with uncertainty. So in the deterministic one, we started off with classical optimization, which is unconstrained optimization using first and second order conditions, using calculus a little bit. Don't freak out, right? You learned this a little bit about how to find a solution to an extreme function. And so we went from that, then we went into linear programming where we constrained it and put some constraints on what the objective function looks like. In fact, it's a linear function. It's a straight line, makes it easier to solve. Then we finally ended up with mixed integer linear programs, which are the workhorse in supply chain management. You will see these all over the place in SC2X when we do network design, production planning, optimized procurement. Mixed integer linear programs are probably the most widely used technique in supply chain management. We then expanded and I showed you some other things like network algorithms. Because again, in transportation, you operate on a network so learning how a network is represented and some different algorithms that are used there are very, very important. You'll see these more in SC1X and later on in your career. We also spent more time talking about algorithms themselves because algorithms are things that you've been using. You just didn't know it at the time and you'll be using more of these. And so we used algorithms in the setting of a network because I wanted to show some of those. So it's kind of a Trojan horse there because I wanted to make sure you understood some network optimization but also wanted to show you how algorithms work. So we talked about shortest path. We I think went through Dijkstra's algorithm. We went through vehicle routing problems, traveling salesman problem. And then we kind of went into heuristics because some of these problems are so big, so complicated even though our computers can do a lot and they're really fast, you still can't solve all these problems. They're NP hard and so they're very large and so heuristics are again commonly used in practice. So I want to make sure you understand when to use heuristics and when to use optimal algorithms. And finally, we finished up with the idea of approximation and estimation because as you go from pure optimization to this approximation, yeah, we'd love to optimize everything but sometimes you don't have the data and it's uncertain. And so you just need to do a rough back of the envelope type of calculation and that's what the approximation is there for. And so we ran through some techniques and some ways of thinking about that and we specifically looked at an approach called continuous approximation for last mile delivery that is widely used and something you should keep in your back pocket. So that was the optimization half which is what I think is one of the funnest pieces to teach. I love that stuff. The second half we dealt with uncertainty and so we introduced you to probability which is again one of the hardest things for people to get their arms around. Usually it takes two or three times to do it. You need to make sure you understand distributions and what that means because you cannot do any inventory management without understanding probability distributions, especially the normal distribution or my favorite, the triangle distribution. I love the triangle distribution. Not everyone likes the triangle distribution but it's a great approximation to other distributions. So understanding what the area under the curve is and what the cumulative probability is, these are things that you're going to see more and more and more especially in SC1X. This will get beat into you because you need to make sure you understand this at a gut level. We talked about then to statistics, we went to point estimation, confidence intervals and then we also did hypothesis testing. These are things that are done if you're implementing a new strategy and you want to test to see did it improve or not. Most in practice they don't necessarily do formal hypothesis testing but if you want to do anything in academia you need to know this really well but the whole structure of being able to say did this new technology, this new methodology, this new facility, did it have an impact looking at the before and after conditions, that's the approach you need to take. So you have those tools now. Then we went into regression which will be widely used. We'll see it in forecasting in SC1X. You'll see it later on in other courses too. It's so easy to do now that it's widely used and so anytime you want to understand some driving forces and their effect on that dependent variable, regression is a nice first step for this and you'll probably use this quite a bit in your career. We finished up with simulation. With simulation we went through Monte Carlo, very simple using a spreadsheet but you can also have very complicated and sophisticated simulations and we did that for the discrete event simulation. We introduced some queuing theory in there as well because that kind of fit that same model. We won't do too much with queuing theory later on. Sometimes you'll see this in different practices. Mainly in manufacturing you'll see a lot of queuing theory but I wanted to make sure you saw it and you don't freak out if someone says an MMQ. So you know what it is. So deterministic, stochastic, we kind of set these together. These are the tools you now have in your tool belt. You've learned more than you think you've realized and hopefully once you do the final you'll come to that conclusion. You will not see any more mathematical techniques throughout the four courses. You've got all the tools there. Now you just have to learn how to use them a little better but you've been kind of taught them on a first take anyway. So that's the course and hopefully you're feeling good about this, about these tools that we gave you, how it fits into larger curriculum. Like I said, you're now equipped to solve every problem that we touch in the remaining four courses. The only exception to that is in SC4X we introduce some newer techniques, machine learning, which will build off of a lot of things we talked about in the second half of this course. But that's only in SC4X. Everything in SC1X, supply chain fundamentals, SC2X, supply chain design, and SC3X, supply chain, what is SC3X? SC3X, the veggie dynamics. Dynamics, sorry. So many SCXs. You've seen before. So you've got those tools. So feel confident and if you like supply chain, that's great because that's what you're going to see for the rest of it. We won't be beating you up on math as much anymore. So that's SC0X, the intro. Anything you want to add, Inma? You've been with them for the last 12 weeks. I think it was a really nice recap of the course. And as you said, even if it was a little bit tough, you're going to be very happy to have these tools in coming courses. And I'm sure you will go back and just review these concepts and these techniques to make sure that you can solve the more realistic problems that will be approaching in one, two, three, and four. Yeah. And throughout your career. Yeah. All right. So now we want to move into the problems for practice, from practice rather. And so we introduced these, gosh, was it last course? One of the SC1X's we introduced it because a lot of students or a lot of learners send us emails and they say, I have this problem. And I saw one of the weaknesses that learners have. Yes, you know all these tools now. You've got your optimization tool in your toolbox. You've got your simulation, your regression, all these different things. But when a problem hits you, they never say, hey, I'm an optimization problem or hey, I'm a regression problem. They're just a problem. And so what we wanted to do with this feature is throughout every course we want to, we always ask for, do you have a real problem that you faced at work? And then we try to talk about it and see, just say, oh, how would we solve this? What would you do? And they're not necessarily really easy to figure out. They're just kind of real problems. And so I wanted to introduce two of those in this and that's what we're going to talk about for the next few minutes and then go to the breakout with this. Okay, so you can download the PDF with the problems in case you want to have them in your computer too from the life event chat. Like Arthur has shared it with you and it's also in the section for the third life event in week 11 of the course. We're going to screen the presentation too right now so you can read it from the screen. You should be seeing now the first problem, problem one, fog lift crew break scheduling. Okay, this came from Victor and he wrote a couple emails on this and they were much more detailed. But essentially the situation was that he is working with a paper company and he's been asked to kind of coordinate the forklift drivers that they're taking a little longer and unplanned breaks during the normal flow of business and so it's not structured so they think it's affecting the performance. And so he's been tasked on a better combination of when breaks should be scheduled for forklift trucks and the drivers at the warehouses. That's kind of the situation, the details. You can think of it, it's a pretty standard situation. Palettes come in from this production line and a forklift takes them to the shuttle and this buffering area at the mill and when the shuttle is full with 15 pallets a driver then takes it on that shuttle to a warehouse where it's loaded, or it's unloaded rather and then truck driver they can return back to the production plant to pick up a new shuttle so they can go back and forth. The shuttle's in the power unit so the thing that the drivers on are separable, they're detachable. Pallets stay around three days at the warehouse and then a truck comes and takes them to the customers. So it's kind of like a staging area so the forklifts are seven of them and they are used for loading shuttles at the plant. There are 42 shuttles that go back and forth and it's a four shuttle drivers, 10 forklifts at the warehouse, that destination, three shifts per day. Just giving you some background on this and the production range is about two and a half pallets a minute and you can do the math to see what that means. It takes two minutes to load a pallet roughly 10 minutes for the shuttle to drive to the warehouse with a full 15 pallets, two minutes to unload and then two minutes to load the pallets when the customer's truck comes. So the question is, you have all this information and you can also, Victor included the distribution of the time and it looks somewhat exponential. Obviously not negative but it has a long right tail for the length of time that it takes to do these loads. And so if you look at this problem and I'll let you, I'll describe the second one first but you wanna probably, I want you to talk about this but how would you approach this? What would you do? This is what's your first step because we wanna go through the process now of you see the problem, identify the problem, identify the best approach. All right, any questions on this one or anything you wanna add? No, I think it's nicely explained. So that's the first problem. Okay, the second problem is container loading and this came from Lance and this is essentially your employed as an analyst at a freight forwarder and this is in Shenzhen. So essentially the freight forwarders don't own assets. They don't own ocean vessels but they are like the middleman between shippers or manufacturers and the carriers shipping things to retailers. So the situation is you've got a lot of plants in China that are shipping things for a common retailer. That could be a Walmart or Target, a Sears, any large retailer. They come from multiple locations and the forwarder's job is to consolidate them onto containers and then they will go to that retailer's distribution centers. So the retailers procure millions of TEUs a year and your jobs consolidate them into full container loads because they come in from plants as two to three pallets, half a container, whatever. You have to mix those together and then you wanna ship them out as full containers. Why? Because you pay for the whole container and you don't wanna ship air. So some details. Retail ship, you could have 10 to 20 containers coming from 100 different origin locations a week in the Shenzhen area, a very hot area. 330 each day, you receive the confirmed volume to be shipped by each retailer from the warehouse and you have about 90 minutes, an hour and a half, to come up with a plan. And that plan is to figure out which ocean shipper to contact and how many containers of which type to lay on. So you have a bunch of input data which is the number of the things you're trying to move in cubic meters, pallets, whatever and you've got to determine how many containers to order. This data can change during the process. It's very dynamic. And so the question is how do I wanna figure this out? What should I do for the approach? And I gave you the four common types of containers that are used, the TEU, the 20 foot standard that holds 26 cubic meters, FEU 40 foot standard that holds 56 cubic meters and the costs are there as well. And then some high cube ones, a 40 and a 45 foot high cube. And so they all have slightly different costs. And so the question is, what approach would you use to solve this? Look at the frequency, look at the input data and what we want you guys to do. And first, is there anything you wanna add to this one? No, no, thank you. Okay, so what we want you to do is to go to your breakout rooms in just a second and talk to each other about how you'd approach this. And there are three things that we want you to really focus on. What data do you think you need? What methodology should you use? And how would you approach using this model? Is it something that's done strategically? Is something daily? How do you see this approach being used? Because everything we've done so far, we've learned kind of an abstract. And now we're gonna say, okay, how can it be used in practice? Because that influences what you can actually do. So what I'd like for you to do now is go to breakout rooms and it's about a quarter after. So I'm gonna go for about 15, 20 minutes. Is that okay? So we'll give you a warning when we're back out, but when you're in your breakout rooms, talk about these two problems, the container loading problem and then the forklift crew scheduling problem. Also, if you have any other questions, send those through in the chat and we'll address those when we come back. When we come back in 20 minutes, we'll talk about each of the breakout, each of the practice problems. Yeah. All right. Yes, so we are aware you don't have time to solve them. Yeah, we don't want you to solve them. Just pick the methodology. That's the weak link here. Pick the methodology and we can talk about it. Yeah, exactly. All right. See you in a few minutes. See you.