Modeling

At a Nextmv tech talk a couple weeks ago, I showed a least absolute deviations (LAD) regression model using OR-Tools. This isn’t new – I pulled the formulation from Rob Vanderbei’s “Local Warming” paper, and I’ve shown similar models at conference talks in the past using other modeling APIs and solvers. There are a couple reasons I keep coming back to this problem. One is that it’s a great example of how to build a machine learning model using an optimization solver....

Note: This post was updated to work with Python 3 and PySCIPOpt. The original version used Python 2 and python-zibopt. It has also been edited for clarity. Back in October of 2011, I started toying with a model for finding magic squares using SCIP. This is a fun modeling exercise and a challenging problem. First one constructs a square matrix of integer-valued variables. from pyscipopt import Model # [...snip...] m = Model() matrix = [] for i in range(size): row = [m....

Note: This post was updated to include an example data file. I thought it might be useful to follow up the last post with another one showing the same examples in R. R provides a function called lm, which is similar in spirit to NumPy’s linalg.lstsq. As you’ll see, lm’s interface is a bit more tuned to the concepts of modeling. We begin by reading in the example CSV into a data frame:...

This post is based on a memo I sent to some former colleagues at the Post. I’ve edited it for use here since it fits well as the second in a series on simple data fitting techniques. If you’re among the many enlightened individuals already using regression analysis, then this post is probably not for you. If you aren’t, then hopefully this provides everything you need to develop rudimentary predictive models that yield surprising levels of accuracy....

Consider this scenario: You run a contest that accepts votes from the general Internet population. In order to encourage user engagement, you record any and all votes into a database over several days, storing nothing more than the competitor voted for, when each vote is cast, and a cookie set on the voter’s computer along with their apparent IP addresses. If a voter already has a recorded cookie set they are denied subsequent votes....

Note: This post was updated to work with Python 3 and PySCIPOpt. The original version used Python 2 and python-zibopt. Data fitting is one of those tasks that everyone should have at least some exposure to. Certainly developers and analysts will benefit from a working knowledge of its fundamentals and their implementations. However, in my own reading I’ve found it difficult to locate good examples that are simple enough to pick up quickly and come with accompanying source code....

One of the difficulties we encounter in solving the Traveling Salesman Problem (TSP) is that, for even a small numer of cities, a complete description of the problem requires a factorial number of constraints. This is apparent in the standard formulation used to teach the TSP to OR students. Consider a set of $n$ cities with the distance from city $i$ to city $j$ denoted $d_{ij}$. We attempt to minimize the total distance of a tour entering and leaving each city exactly once....

Uncapacitated Lot Sizing (ULS) is a classic OR problem that seeks to minimize the cost of satisfying known demand for a product over time. Demand is subject to varying costs for production, set-up, and storage of the product. Technically, it is a mixed binary integer linear program – the key point separating it from the world of linear optimization being that production cannot occur during any period without paying that period’s fixed costs for set-up....