Operational Recollections

I'm not a CS major of any sort. In fact, I only took 2 programming classes in college. Truth be told...I initially did not like it after my freshmen intro course. But by my junior year, I saw the practicality of it and decided to give it another go.

Starting out, I did some pretty simple things. My freshmen intro class went over the fundamentals in C++. The class I took in my junior year went over Databases and Visual Basic.NET. I was "skilled" enough to create simple desktop applications on VB.NET. The following summer at my internship, I built a tracking application in VB.NET that connected to a MS Access database.

As an industrial engineer, I was taught to think and analyze data in a certain way to make decisions.I often thought about how I can "automate" the data-collecting process and display the conclusions in real-time. This would allow users to spend more time "doing" and less time "thinking". I was fascinated by what I can automate. This was pretty much why I took my first job out of college as an Industrial Engineer in the Decision Support Systems group. I was fortunate enough to have such amazing mentors and teammates. Not only were they able to think about business operations in a strategic level, they were also able to develop software solutions to implement their plans. In the semiconductor manufacturing world, this was next-level shit. We were doing things that only a few companies in the world could even dream of doing. As the new kid, I learned and learned and learned and learned. This was how I got into web programming in ASP.NET, C#, HTML, CSS, and JavaScript. 

But after I quit and took a position at another company, I began to plateau. I wasn't learning new technology. I wasn't the new guy anymore. I think at some point, I became complacent with what I can do. Frustrated by my lack of growth, I finally signed up for a course in HTML5, CSS3, and Javascript. In my opinion, I felt that I could improve in those areas. I never really had formal web dev training so this was very insightful for me. It reinforced a lot of what I already knew but it also exposed me to some of the industry's best practices. 

I'm very eager to begin coding away again. I don't really care what I'm building. I just want to do something and put myself out there. To brush up, I'm currently using the following resources: 

- https://dash.generalassemb.ly/projects

- http://tryruby.org/levels/1/challenges/0

- http://www.codewars.com/

- http://www.teamtreehouse.com/

- http://www.hackdesign.org/lessons

- http://www.bentobox.io/

"Grit is passion and perseverance for long-term goals. Grit is having stamina. Grit is sticking with your future, day in, day out, not just for the week, not just for the month, but for years, and working really hard to make that future a reality. Grit is living life like it's a marathon, not a sprint"

There are many forecasting methods and one that interested me recently is forecasting based on seasonal trends. A seasonal series repeats at least every 3 periods. Seasonality is typically represented by a multiplier.

If the multiple for Monday = 1.20 then the demand on Monday is 20% above the average demand. Conversely, if the multiplier for Tuesday = 0.75, then the demand on Tuesday is 25% lower than the average demand. 

To calculate the factors of stationary series, you need to follow these steps:

1. Calculate the mean of the entire sample data. 

2. Divide each data-point by the mean to get the seasonal factors of each data-point. 

I have recently quit my job as an Industrial Engineer in the semiconductor manufacturing industry and joined another company. Before leaving, however, I had my pick of 3 great companies. What was interesting was that all three companies were in completely different industries (consumer electronics, semiconductor manufacturing, and healthcare IT). I think that's the beauty of being an Industrial Engineer. We are generalists by nature. 

My new role requires me to innovate and improve in a big company...which is not the easiest thing in the world. Guy Kawasaki put it nicely: "Creating a new product or service inside such a beast is not necessarily easier; the challenges are just different." 

So how do you do it? In his book (Art of the Start), he mentioned a few key points that I thought were applicable to me:

1) Put the company first - Don't try to change, disrupt, or innovate simply to stand out above the rest. Do it so the company can benefit from it. You'll get more buy-in from upper management when they notice your project. 

2) Stay under the radar - Don't go flaunting your project ideas if you know there are people who may feel threatened by it. Innovation and change are dangerous to some people. Only reveal your project when you know it is already too big to shut down. 

3) Find a godfather - Simply put...find someone who is senior enough to support your project. This person needs to have a certain level of influence and authority in the company. 

There are three typical outbound distribution methods:

Direct Shipment

Warehouse

Cross-Docking

Direct Shipment - Materials are transported directly from supplier to retailer or manufacturer to retailer. This method takes a distribution center (D.C) completely out of the picture. The major advantages to utilizing this strategy are the reduction in lead times and the expense savings of not operating a D.C. The direct strategy allows manufacturers or suppliers to directly ship products to their destination. 

However, the major disadvantages to this method are the increase in transportation costs and the loss of risk-pooling effects. Without a central distribution center, the firms in the supply-chain would be susceptible to high demand variability. Because of the potential high variability, firms may have to ship more frequently as demand sways thus increasing the transportation costs. 

Retail firms with strong buyer power may commonly exercise this strategy as it requires fully loaded trucks. 

Warehouse - Most common and traditional method for distribution. A warehouse keeps the inventory on hand and provides products whenever retailers need to restock. With this method, companies take advantage of the ability to risk-pool products and aggregate variability. They can also reduce the cost of inbound transportation but as a result, the lead-time may increase. 

Cross-Docking - This strategy utilizes warehouses not as an inventory storage but more as inventory "check-points". Manufacturers would ship products to the warehouses and the warehouses would then send the products to the retailers as soon as possible. This reduces a large portion of lead-time and inventory holding costs. However, the cross-docking methodology requires a large capital investment. Firms usually use this strategy if there is a very large volume of goods being moved each day. In order for cross-docking to be effective, the distributors must have a fleet of readily available methods of transportation as well as a tight-knitted connection with the retailers and suppliers. This is where information and sharing of information are so important. When cross-docking is properly implemented, the supply-chain runs like a symphony (see Walmart's distribution strategy). 

So what factors should we consider when deciding on a particular strategy? Well, who said we have to use them exclusively? Many companies employ a mix of strategies and ideas. Firms deciding on their distribution strategy should consider the following:

Came across this great piece by Jake Levine. He's essentially saying the best way to learn programming isn't to go through tutorials and mindless exercises but to actually create something. I couldn't agree more. Having a solution or final product in your head gives you that passion to spend hours and hours figuring out syntax errors, design flaws, and proper logic. I know I get a kick out of programming if I'm building something that I know has value. It gives you a goal that you're constantly striving to reach for. 

Imagine a supply chain that consists of the following:

Supplier 

Manufacturer

Distributor

Wholesaler

Retailer

Materials flow from #1 to #5 but information flows from #5 to #1. As retailer sells more and more widgets, they need to restock their inventory. So to do that, they place an an order to the wholesaler for a shipment of widgets. The wholesaler ships these widgets to the retailer but the wholesaler eventually needs to restock. So the wholesaler then communicates with the distributor to obtain more widgets. In order for the distributor to replenish their inventory, they place an order with the manufacturer. The manufacturer needs more raw materials to create these widgets so they place an order with the suppliers. 

Keep in mind that while all this is happening, each individual node of the supply chain is managing their own inventories. The retailer might forecast the customer demand will be higher than average. So he places a higher than average order to the wholesaler. The wholesaler sees that demand just skyrocketed and is rushing to get enough widgets out to the retailer. To prevent this in the future, the wholesaler keeps a safety stock of widgets for the "just in case" scenarios. Well, the wholesaler needs more widgets so he places an order to the distributor. And so on and so forth up the supply chain. 

Because each node is managing their own inventory, they only have a vague idea of what is happening with customer demand. They forecast and build safety stock based on someone else's forecasts. The variability in demand is forcing each node to carry more and more safety stock and inventory buffer. As we move upstream, we see that the safety stock gets bigger and bigger because the uncertainty gets bigger and bigger. 

Lead time also plays a huge role. The longer the lead times between each node, the more safety stock we need to carry. Some companies may also order in batches to take advantage of discounted rates and reduced transportation costs. The fluctuation of prices can lead some retailers to "stock up" when prices are low. 

In summary, the following are main contributors to the increased variability:

Demand Forecasting

Lead Time

Batch Ordering

Price Fluctuations

Inflated Orders

So how do we reduce the bullwhip effect? 

Inventory appears in multiple stages of the supply chain:

Incoming raw materials from suppliers

WIP (Work in Process) in manufacturing

Finished goods

With all the craze about lean manufacturing and JIT distribution, inventory has gotten a bad rep lately. While it is true that excessive inventory is bad, there are also several reasons why we should hold inventory. 

Companies hold inventory as a hedge against demand variability. High customer demand variability in the recent years has increased due to many products with short life cycles and the overload of competition in the markets. The short life cycle of a product makes it hard for analyst and engineers to accurately use historical information as a basis for decision making. The large number of competing products makes it difficult to predict demand in general. 

Uncertainty and variability from the suppliers. Because many manufacturers rely on outside vendors for their raw materials, they are affected whenever the suppliers cannot meet their quality and/or schedules. This reliance on outside suppliers makes it hard to completely reorganize and improve the overall supply chain.

Firms can potentially use the economies of scale to their advantage. Ordering in batches, transporting in truckloads, and holding in bundles could give them discount rates. 

Consider this.

In scenario A, our company has 2 warehouses to serve 2 separate markets respectively. We'll call the first warehouse, WH-A and the second warehouse, WH-B. 

WH-A solely distributes to the Michigan market while WH-B solely distributes to the Texas market. This is a case of a decentralized distribution system. 

In scenario B, our company has 1 warehouse located between Michigan and Texas. This warehouse will serve both the Michigan and Texas markets. This is a case of a centralized distribution system. 

The obvious benefits in scenario A (decentralized system) are the speedy delivery times, faster response time, closer and simpler distribution network, and ability to serve market demand. On the other hand, the benefits in scenario B (centralized system) are the lower safety stock inventory and lower overhead costs.

As we increase the number of warehouses, we reduce the outbound (warehouse --> retailers) transportation costs but we also increase the inbound (manufacturing --> warehouses) transportation costs. 

// Side-bar here

The Coefficient of Variation (CV) = (Standard Deviation / Average Demand)

CV measures the variability in relation to the average demand. It is a normalized measure relative to the mean. Standard deviation simply gives you an absolute measure of variability across the mean. 

//

What does all this have to do with Risk Pooling? Well, we are able to reduce the variability in demand if we can consolidate demand across multiple locations. The rationale here is that we can potentially offset low demand in Michigan by having high demand in Texas. Vice-versa. This allows the company to have a much lower "safety-stock" of inventory. 

The higher the CV, the more variability there is in the market demand. In a simplified high level example, the centralized system would be beneficial if the CV is high. If the CV is low, the benefits of a centralized system would not be as obvious. 

Victor Cheng's "Profitability" Framework. Start at a high level and dig deeper:

Profits = Revenue - Cost

Revenue = Volume * Price

Victor Cheng's "Business Situation" Framework.

Customers

Product

Company