ICP In Practice

This post explores the iterative constrained pathways rule ensemble (ICPRE) method introduced in an earlier post using the Titanic dataset popularized by Kaggle [1]. The purpose of the text is to introduce the features and explore the behavior of the library.

Some of the code snippets in this post are shortened for brevity sake. To obtain the full source and data, please see the ICPExamples GitHub page [2].

The Iterative Constrained Pathways Optimizer

Many optimization methods seek an optimal parameter set with regard to error or likelihood. Such a solution is most desirable in many regards. However, when the broader context of a problem is included, the indisputable superiority of the optimum frequently becomes less clear. This context often includes other guidelines and restrictions that may limit the usefulness of solutions lacking certain properties. Unfortunately, typical loss criteria can rarely take these into account.

This blog post presents a method that abandons the quest for optimality and instead focuses on better satisfying the broader context of a problem. It describes a method that does not attempt to find the minimum, but instead simply tries to get closer to it while respecting imposed constraints. This blog post describes the iterative constrained pathways optimizer.

Simulated Genetic Drift in Populations

This video shows a simulation of genetic drift in a synthetically generated population as a result of sexual reproduction. A fixed population size is divided into different distinct sub-populations with differing allele frequencies. Starting with an initial population, 80 simulated epochs are passed in which each population member is replaced via sexual reproduction from two randomly selected parents. Pairs of parents are chosen such that the probability they come from the same sub-population is higher than the probability that they from come different sub-populations.

Figure 1: Simulated Genetic Drift in Populations

In the animation, the high-dimensional allele features for each population member are represented in two dimensions using PCA. The size and color of each marker encodes information about sub-population makeup and may be used to help distinguish highly mixed samples.

Although the data is entirely synthetic, it demonstrates the middling affect that cross-breeding has on the genetic makeup of the overall population. As the number of sub-populations grows and the overall population size remains fixed, the impact of cross-breeding is intensified and the population as a whole converges more rapidly.

CoV-2 Genome Explorer

At the time of this writing, NCBI Virus has over 40,000 complete CoV-2 sequences available. I made an interactive webpage to help visualize a random subset of these genomes using principal component analysis.

Figure 1: CoV-2 Explorer Interface

The x, y, size, and color axes respectively represent the first four principal axes. The plot markers identify viruses with unique nucleotide sequences in the GU280_gp02 (spike glycoprotein) gene region. The range sliders can be used to visualize shift in the viral genomes over time. The bar chart in the bottom right counts the frequency of SNP mutations in the sequences present in the specified date range.

Ancestry Determination Part 3: openSNP Data Evaluation

The website openSNP allows users to share and discuss genetic information [1]. The purpose of the site is perhaps best summarized by its bio on Twitter: “crowdsourcing genome wide association studies” [2]. This post uses the machine learning and statistical techniques developed in previous blog entries to analyze the ancestry of openSNP members who have genotyping data associated to their accounts. By comparing the derived results with the self-reported answers from openSNP users, the accuracy of these methods on real-world data is assessed.

Identifying Family Relationships using Genetic Similarity Measures

In general, are parents more genetically similar to their children or to their siblings? Why do some siblings look alike while others look completely different? How can the genetic similarity between two individuals be computed? This post uses biostatistics to explore genetic similarity and tries to answer these and other questions.

Decorrelation Redux

Consider the typical statement of the least squares problem.

,

where is the m x n data matrix, is the n x 1 vector of regression coefficients, and is the m x 1 vector of target values.

Analysis of the Preliminary 2020 Election Results

The 2020 general election in the U.S.A. is interesting in several regards. Perhaps chief among these is the influence of the COVID-19 pandemic on voting patterns and the subsequent shift to more reliance on mail-in voting. This post presents analysis of the preliminary election results and attempts to explain some of the observed trends.

Analysis of the 2016 US General Election

Counties within the United States vary substantially along a number of demographic and socioeconomic axes. These factors explain much of the variation observed throughout the US with respect to election and polling trends. This post attempts to untangle some of this complexity to give insight into the factors that influence voter behavior and other broader national trends.

The United States by County

This post explores the composition of the United States using choropleth charts. The statistics and other values represented are as of the year 2016. Percentage changes in the number of democratic and republican voters are computed between the years 2008 and 2016.