Using the popular bird variety data to demonstrate nearest neighbors classification
Introduction
This article uses fictional data, previously generated using the code in an earlier article to illustrate the k-nearest neighbors classification algorithm. Readers can use this article as a cookbook for executing classification algorithms with the k-nearest neighbors algorithms.
Both a Jupyter notebook and a YouTube instructional video accompany this article. I placed links to these additional resources at the end of the article.
The Algorithm
This algorithm, k-nearest neighbors, is one of the simplest supervised machine learning algorithms available for classification. Frequently, the k-nearest neighbors algorithm performs as well as many other more sophisticated machine learning algorithm options.
The underlying principles at…
Common, or not so common career paths Towards Data Science. How teachers, faculty, accountants, engineers, artists and others get into data science.
Introduction
There are many ways to pursue a career in data science. This article covers six possibilities. Of all the possibilities two things to keep in mind. One, there are more than seven paths. Two, there is no right or wrong way to go about it.
From my frequent and ongoing discussions with other data professionals, I wanted to share a summary of common paths towards data science.
For example, I spent most of my career working in education. In my career, I have worked in the classroom. I taught abroad. Also, I moved into education administration before I finally transitioned…
We automate data collection to save time, reduce error, add rigor, and promote reproducibility
Introduction & Method
After a recent article on the topic of sourcing federal data, in which I show how to use Python to automate the process of getting data from the US Department of Education (US DOE) and then assembling that data into a panel data set, I started getting questions.
Why would you go through the trouble of writing code for this? Wouldn’t it be faster to just download the files and then use point and click to assemble the data?
The answer is, it depends. I did an experiment. In executing this experiment I recorded myself as I used point and…
Have you considered a presentation on this topic (spider and radar plots) for Stata 2021 conference?
A brief inventory of tools that make fictional data
Since writing about this topic earlier, a handful of folks throughout the community have shared with me their own picks for tools that generate fictional data. I evaluated three tools to see how well they can produce the fictional data I previously wrote about. Here are the results of how I evaluated these tools.
1) Faker — Gets very close.
2) GenerateData.com — Gets close.
3) Mockaroo — Gets very close.
4) On Your Own — Perfect Match.
In no particular order here they are. Below I write a bit about each. I evaluate each on a three-point scale to…
When you need data for testing, training, demonstration, or other purposes. Make your own!
About once a week I get (or see online) a question from a fellow data scientist or aspiring data scientist. Where can I get a data set to play with? Or, I’m looking for an interesting data set to learn with, any suggestions?
There are plenty of interesting data sets out and about. But why not make your own data? Making your own fictional data is also a useful skill when you need data for testing or demonstration purposes. This article will show you how to generate fictional data (this is one set of many methods).
At the bottom of…
A demonstration with data from the US Department of Education, how to use distance measures to specify college comparison groups
Introduction
Imagine a scenario in which a college or university receives criticism in the form of negative print, online, and social media attention. Suppose that attention focuses on the institution’s undergraduate application fee rates.
How can the institution respond? If your thought on responding to this media attention involved comparing the institution, to other similar institutions, this article is for you. This negative media attention hypothetical provides a case study for applied use of distance measures below. Institutions can use these distance measures to identify meaningful comparison groups. …
How I’m gathering feedback from students, amid our four-plus semester-long pandemic
Introduction
Getting feedback from students has always been a priority of mine. I collect feedback early and often. My practice has followed a specific format that asks open questions.
- What is something that is going well so far? What, if anything, would you like to see more of?
- What, if anything, isn’t going well so far? What would you like to see less of?
- What questions about the course requirements, assignments, and expectations do you have?
- What questions about the course material do you have?
I’m also finding, and I think this is perhaps due to a lack of in-person interactions…
Proposed: The goal of data-driven organizations, startups and others, is to generate new knowledge.
Introduction
A well-designed data-driven culture is optimized for transforming data into new knowledge. Organizational leaders own responsibility for enabling that optimization. Organizational members own responsibility for engineering that optimization. Among the many valuable products any startup, or other organization, can offer, is new knowledge.
Knowledge, especially new knowledge, is the goal. Before the data-driven leader can build a data-driven culture, we have to know what a data-driven culture is. The leader, in collaboration with organizational members, must decide about what data-driven culture means. This guide offers definitions for the term “data-driven culture” and its constituent terms.
Data-driven leaders, their organizational, along…
