# Is Hillier F. Introductory to Operations Research a good book for a data analyst interested in Operation Research field?

Is Hillier F. Introduction to Operations Research a good book for a data analyst? I was particularly interested in Operations Research especially when I was reading about the history about this area:

This field was created to the Necessity of Applying Maths in the World War I and II.

I researched about books here: Recommended books/materials for practical applications of Operations Research in industry but it's a broad recommendation for me and I am starting to go deeper, however I already know linear algebra, calculus, differential equations, statistics, Excel, Python, and other analytical tools for handle data. I am taking an Engineering and Math Degree so maybe that could help me.

I don't want a book that only worth for me for tutoring. I want a book with real applications for possible get a job in some interesting industry related to OR. So if you would recommend me one better than Hillier for a start and take shortcuts will help me a lot.

Edit: I saw a lot of software for OR in this question here, so I would like take that software and language that I already know (Python, Excel).

Having previously worked in software engineering, then trained in mathematics (stats & OR), and since worked in OR for ~3 years, I have a strong opinion on this.

If you want to work in OR then it is expected that you are familiar with the topics in the book including some theory behind LP, MIP, and so on. In practice, the real difficulty in tackling a real world problem is in framing the problem in such a way that those methods can even be used in the first place. This is a different skillset. Also recognizing when a problem is too complicated for these methods to be used and it is necessary to resort to metaheuristics and in some case custom heuristics.

Having attempted many of the methods in that book including LP, MILP, MDP, graph algorithms, SA, GA, Decision trees, Markov chains, some queuing models, OptQuest..... my conclusion is that the clean mathematical methods tend not to fit many real world problems, at least not the really complicated problems, which is what I have faced almost exclusively in my work.

Rather, the only things which "worked" (by worked I mean - actually solved the real, full problem, and were actually released to be used in practice... for real... in the actual real world) were heuristics, meta-heuristics, simulation based optimisation, and statistical analysis of simulation output. To the point now where I don't bother trying the nice clean mathematical models on any big problem because it's a waste of time.

The book, unfortunately, seems to talk about the nice mathematical models and theoretical side. It is necessary to be able to talk in that language to operate in the field. But don't be fooled - the success criteria for many projects in this domain is "I got a paper published in a good journal", which in my experience is negatively correlated with "I got it to solve the real, full problem and got people to actually use it".

For that reason no I don't think this book will introduce you to many practical techniques you can apply to real world problems. However, it will equip you with a useful language and set of ideas. And you will know what the options are for computing solutions if you know that it can be reasonable represented in vairous formed, like LP, MDP, parameteric black box function, etc.

• This is the real answer that I was looking for a skeptical answer that respect the complex world that we living. In the statistics field usually we said: "All models are wrong, but some are useful" this is the mainly problem with "complex models". Despite this @Brendan are saying that knowing this models and use this for the right time, right place and in the right moment.” will be useful. thk u for shedding a light on my question, and all of you for take your time answering and recommending books for beginners in the OP field. Commented Sep 10, 2021 at 2:46

You can consider going through below blogs:
i) Erwin Kalvelagen's blog
ii) Prof Paul Rubin's blog (https://orinanobworld.blogspot.com/)

Interesting things about above blogs is that you can see different applications, their implementations and practical issues while handling them. Another good course is by Prof. Pascal Van on Coursera (https://www.coursera.org/learn/discrete-optimization). You can check that course. Also can follow tutorials/webinars from different solvers (mainly CPLEX, GUROBI, LOCALSOLVER, FICO, SAS etc..)

• Thanks for the endorsement. I just want to point out that Erwin's blog and mine are largely limited to discrete optimization. It would also be good for someone new to OR to find similar resources in other domains, particularly simulation and stochastic processes.
– prubin
Commented Sep 8, 2021 at 15:21

For an idea of industrial applications of OR write large (minus the gory technical details), you might want to look at a few INFORMS publications.

1. Analytics (aimed at the general public) and OR/MS Today (for INFORMS membership) contain articles about implementations.
2. The INFORMS Journal of Applied Analytics (formerly known as Interfaces) contains primarily articles on real-world implementations of OR methods.

Another place to look is software vendor web sites, which may display synopses of real-world solutions using their product. A quick search for "simulation case studies" yielded a page on the AnyLogic web site.

I found it quite useful to start (social scientist with background in stats). I had mostly read more complicated papers using linear programming in geographical allocation contexts, and was totally baffled by it. (Relative to stats, in retrospect people using X/Y as decision variables and Greek letters as fixed values is where quite a bit of my confusion came from.)

Scooped up a Hillier book from the library, went through the introductory formulating of problems, and then had a much better starting point to understand the more complicated use cases. (Don't quiz me on dynamic programming though.)

The copy I used had problems in Excel, but it was not much work to follow along implementing them myself in python (I use pulp, but could use whatever library you want really).

The step from toy examples to real world applications is IMO not a big one. That is on you to formulate the real world problem in the context of your business. Pretty much all of the machine learning models I use in production have some sort of decision analysis piped on the end in practice that if I have particular constraints I use linear programming for (this example is common at my workplace).

Hillier and Lieberman is a good read and solid reference to get the "big picture" of many techniques. You may also want to look into Operations Research: Applications and Algorithms by Winston.