So my wife and I are buying some practical things and other to decorate it. She’s managing some things and I’m managing others, so we can better split the work and be more efficient. And we’re buying in different platforms too.

I don’t want to bias you so I won’t say those platforms, but the thing is last Monday I bought (let’s say at Platform A) one soap dish and two flowerpots and they arrived just one day after that, on Tuesday, and she bought (at Platform B) some cushions for the couch and one basket for dirty clothes two weeks ago and never arrived. In fact, she had to cancel the order and get the money back.

But why is this happening?

Why did I order something and magically appeared on my doorstep the next day, but my wife didn’t get her order?

Is it because the shelves at Platform A are always stocked with everything users need and Platform B’s shelves are always empty? Because some differences in the producers? Or anything else?

The answer lies in the complex world of supply chains, the intricate network that brings goods from producers to consumers.

But lately, these supply chains are facing some serious challenges.

Rising fuel costs, labor shortages, a tangled mess of outdated systems…

This is the reality for many supply chains today. They're fragmented, slow to digitize, and struggling to keep up with ever-increasing customer demands.

Delays, inefficiencies, and unhappy customers become the villains in this story.

Since we have big players in the table, are they giving us anything to solve this issue?

Today in Feasible we’ll see:

• 🗺️ What’s the Google’s Supply Chain strategy map.

• 🌐 A job offer from Microsoft that tells us they are working in supply chains… In the digital world.

• 🚛 A Spanish startup raising some money thanks to solving optimization problems in a daily basis.

Are you ready for this? Let’s go for it!

🗺️ Supply Chain from the lens of Google

Recognizing these supply chain challenges, major tech players are bringing their AI/ML muscle to the fight.

And Google isn't just in the business of search engines anymore.

They're using their expertise in AI and machine learning to develop innovative solutions for supply chain management.

One of their key strategies involves utilizing digital twins. This allows Google to model different scenarios, predict potential problems, and identify opportunities for improvement.

Google is leveraging its strengths in AI/ML, cloud computing, and access to rich mapping data to build a suite of supply chain solutions:

Read more

Not just liked, but loved.

It's one of those things that I read and it makes me happy.

Walmart, probably one of the companies with the most experience in the world of Operations Research, unveiled to the world its Route Optimization tool.

Just by the name, you can imagine what it's about.

In the true style of UPS, Walmart provides concrete data. With this tool, they have saved:

• More than 110,000 routes.

• More than 48 million kilometers.

• More than 42,000 tons of CO2.

So, if this is not enough incentive... Just so you know, in this post from Feasible, I tell you about this and an additional little surprise that I got from Walmart.

Let's go for it!

🚚 Route Optimization

One of my first posts in this newsletter was about the UPS case. It was the fourth post and a lot has happened since then, so I'll leave it here in case you don't remember:

The UPS use case might be one of the most iconic within the world of Operations Research.

It provides very concrete data on everything that their ORION tool saved them. It took several years to build, yes, but it was worth it: from the moment it was deployed, it saved the company 50 million dollars.

Now it seems like it's Walmart's turn to strut their stuff with their numbers thanks to Operations Research.

I have to say that for as long as I can remember, that is, for about 10 years now, I've seen Walmart in a very positive light: they post job offers related to OR and occasionally publish on their blog some of the work they do.

With this tool, Route Optimization, Walmart continues its mission to make the supply chain more efficient, also reducing CO2 emissions into the atmosphere. They made it for themselves and now offer it to the rest of the world. With this tool, you can:

Read more

Amazon plays a key role here: it’s one of the companies that make those emissions and, at the same time, developed a tool that provides a means to strategically deploy electric vehicle (EV) charging infrastructure across Europe and support the logistics industry's shift towards electric trucks.

Alongside vehicle electrification, improving fuel efficiency, and promoting sustainable transportation modes, initiatives like CHALET underscore the critical need for innovative solutions and collaborative efforts to drive the decarbonization of transportation and help Europe reach its environmental targets.

Today in Feasible you’ll see:

• Ways of going climate-neutral by 2050 in European Union

• Why electrifying vehicles is the biggest step on that objective

• How can Operations Research help on that quest by setting up an strategic plan for the next 30 years

Let’s go for it!

🌱 Climate-neutral by 2050

The European Union has an ambitious goal by 2050: becoming climate neutral.

This means achieving a balance between the amount of greenhouse gas emissions produced and the amount removed from the atmosphere. By 2050, the EU aims to ensure that any emissions released are fully offset by absorbing an equivalent amount from the atmosphere, effectively bringing net emissions to zero.

How can it be achieved? Mainly by two specific actions:

• Dramatic reductions in emissions. This entails significantly cutting emissions across all major sectors, including energy, transportation, industry, agriculture, and buildings. It involves transitioning to renewable energy sources, enhancing energy efficiency, electrifying transportation, and adopting sustainable practices across industries.

• Carbon sequestration. Beyond reducing emissions, the strategy involves enhancing and expanding natural carbon sinks, such as forests and wetlands, which absorb CO2 from the atmosphere. It also includes exploring technological solutions for carbon capture and storage (CCS) and carbon capture and utilization (CCU) to remove CO2 directly from the atmosphere or prevent it from reaching the atmosphere from emission sources.

This goal is central to the European Green Deal, which outlines the EU's strategy to become the world's first climate-neutral continent and is part of the EU's commitment to the Paris Agreement, an international treaty on climate change.

Achieving climate neutrality involves a comprehensive approach that also includes other initiatives like circular economy to reduce waste, policy and regulatory measures to drive the transition, and international cooperation to promote global climate action, among others.

Look.

The EU is doing a lot of efforts around this. But if the trend continues, I don’t know if those ambitious goals will be achieved. There’s been a cut in the emissions of around 30% in the last 30 years, going from 4.66 millions of kilotons in 1991 to 3.24 millions of kilotons in 2021:

So following the same amount of reduction for the next 30 years, that same 30% reduction means we’ll still be emiting 2.27 millions of kilotons by 2051, one year after the objective. Will that be enough?

Time will tell.

But if I had to focus on something, I would order all the sectors contributing to greenhouse gas emissiosn and select the ones with high values. You can find several statistics and all of them say exactly the same thing:

Read more

Of course, having a runny nose, coughing, breathing poorly, feeling more tired than usual, or finding it very difficult to climb the stairs in my house does not prevent me from writing a new post of Feasible.

Today, I also bring you a topic that I've wanted to write about for a long time.

If you had to choose between solving a problem in a practical way or solving it optimally, which would you choose?

I'm clear about it.

Amazon is clear about it too.

Jeffrey D. Camm, a professor in the Operations Research department at the University of Cincinnati, also is.

We all choose the first option. Why? Well, for 3 very powerful reasons:

1. You invest less time in developing the solution.

2. It's easier to solve computationally speaking.

3. The solutions tend to be more flexible and adaptable to the real world.

I won't go on any longer... Let's get to it!

📦 How Amazon groups orders in its deliveries

The other day I made a couple of orders on Amazon. I placed them separately, but they were scheduled to be delivered on the same day.

I wondered if they would arrive at different times, but no, they arrived together. If they had arrived in different deliveries, it would have been a hassle. Actually, two hassles.

A hassle for me, because I would have to be at home waiting for those two orders to arrive.

A hassle for Amazon, because they would have to come twice when they could have come just once.

As I say, everything arrived in a single order. It's not a coincidence: they've done it on purpose.

I won't tell you the technical details because you can read them here1, but I would like to highlight Amazon's way of working.

Simplifying a problem means losing optimality, but gaining manageability

Among many other things, the article I linked above 👆🏻 mentions that to efficiently group orders, they spent several years of development and many different techniques such as:

Read more

• How to improve an initial solution with local search methods

• The market share of Mercadona and the problems that arise around its warehouses, the famous beehives

Ready, set...

I recently created a database with over 70 Operational Research resources.

FREE.

And it includes things like:

👨‍👩‍👧‍👧 Top communities

🏀 The best courses

🎙️ Podcasts where you learn a lot

⚛️ Useful software for Operational Research

➕ And much more...

I would infinitely appreciate it if you could give it 5 stars ⭐⭐⭐⭐⭐ if you find it useful.

You can get it here.

⬆️ We already have an initial solution, shall we improve it?

In case you don't remember, a couple of weeks ago, we saw how to build an initial solution. In fact, we explored 5 different methods:

All of them are quite fast, and they all have the same problem: the solution you'll find for your problem is... well, I don't like to say these things, but yes, it's not good.

But it's a starting point.

Let's say it's like when you go for a hike in the mountains, and your goal is to reach the summit of a mountain. So, with this, what we've done is parked the car right where the trail begins. You still have a way to go, but you're in a better position than when you left home. And what do you do to start the journey?

Get moving.

In optimization, a move is a change in the solution that you determine. This move defines a specific neighborhood; and it's called a neighborhood because the solutions found from a solution and a type of move are close to that solution. Moving through the solution space, finding better solutions (in one or several neighborhoods) until you can't find a better solution anymore is what's known as local search.

This local search evaluates each solution it finds, and if it's better than the current one, it keeps it. When no better solutions are found, it reaches what we know as a local optimum.

If you notice, this concept of local search is based on what is probably the oldest procedure in optimization: trial and error. You try a new solution with a given move, and if it's better, you stick with it, but if not, you keep looking.

For example, it's very typical in vehicle routing problems to make moves like this:

In which initially we have the connection b-e and c-f, and we end up linking b-c and e-f to improve the solution. In fact, this specific move has its own name and is called 2-opt, although I repeat: you can define the types of moves you want.

Let's continue.

When exploring the neighborhood defined by a move, there are two classical strategies:

Read more

They are building it anew.

I have been documenting the process with photos for months. I may seem like a retiree, I know, but I find the pace they maintain fascinating, and above all, the efficiency of the workers.

The day they hand over the apartments, they will surely throw a party. Just like when a company goes public.

All of this sums up well what I want to tell you today in Feasible:

• How to build solutions to your optimization problem

• The IPO of Instacart

Let's go for it!

🏗️ 5 Ways to Build a Solution for Your Optimization Problem

The other day, I talked to you about how I would manually solve the VRP problem, and the first step was to build a solution. You don't need to remember it because I'll leave it here:

The important part was the beginning of the second section:

Start with an initial solution. For example, I would assign each customer to a random vehicle.

Okay, how do we build that initial solution? There are several methods, and today I'll show you the 5 that are possibly the most commonly used:

Read more

Moreover, when it comes to a league as large and complex as the NFL (National Football League), with 32 teams, 17 weeks of regular season, and multiple factors to consider, the process becomes extraordinarily intricate. Holiday dates, stadium availability, TV contracts, and much more come into play.

Today on Feasible, we will explore how the NFL tackles this logistical puzzle and the fascinating challenges it faces in creating a fair and appealing schedule for fans.

4000 AWS machines work tirelessly for three months to bring an exciting season to life.

Let's go for it!

🏈 Starting the year on the right foot

How easy is it to plan a schedule for a sports league, right?

1. You pick a team

2. Week after week in the first half of the league, you assign them an opponent under the condition that:

   1. Opponents cannot be repeated

   2. Each week they play once at home and once away

3. You replicate the same schedule for the second half of the league, swapping the home and away games

4. Once planned, you move on to the next team, repeating the process from step 1, as long as there are teams left to plan

Read more

I'm excited to announce that I'm launching a new type of content for Feasible: medium-form posts. And they even have a name: OR in action. These posts will be shorter and more concise than my regular long-form articles, but they will still be packed with valuable information about OR, Data Science, and AI.

In OR in action I'll be covering a wide range of topics, including:

• ✅ Sucess stories around OR in businesses

• 💥 New OR techniques and algorithms

• 📰 Recent OR conferences and papers

• 📚 OR books, blogs, and libraries

• 🤖 AI applied to OR (and vice-versa)

• 🔢 Solvers

• 💭 Concepts in OR

• 🖊️ Specific mathematical modelling things

I'll also be sharing more of my own personal experiences and thoughts on OR, and I'll be looking to engage with you all more directly in the comments.

I hope you'll enjoy this new type of content, and I look forward to hearing your feedback, which is a gift.

And it’s already starting… So let’s go for it! Today we’ll see:

• How to start applying OR in any business

• ChatGPT for solving optimization problems

🤔 How to start applying Operations Research in any business

Last Friday I shared this tweet:

Read more