A Non-Technical Introduction to Machine Learning

Why should I read this, and what will I learn?

  1. Understand the basic logical framework of machine learning (ML).
  2. Be able to define important relevant terms and concepts that anyone interested in this field should know. These terms are highlighted in boldface.
  3. Know which high-level decisions go into building statistical models, and understand some of the implications of these decisions.
  4. Be able to better analyze the question of when we should use the results of ML to make big decisions, such as determining public policy.

What is machine learning?

  1. Prediction: make predictions about the future based on data about the past
  2. Inference: discover patterns in data

What’s a statistical model?

A framework for understanding ML:

Example of structured data. Each row is a person (a unique observation we’ve made), and each column is a different measurement (called a feature or predictor) of that person.
  1. Regression models, which are used when the response variable (i.e. the variable that you’re predicting) is continuous. For example, height, age, and income are all continuous. That is, they can be placed and ordered on a number line.
  2. Classification models, which are used for categorical data — that is, data that doesn’t have a numerical ordering. For example, you may want to predict, based on an image of a flower, the species of that flower. Or you may want to predict whether a student is a psychology major or a math major.

A common regression model: simple linear regression

Structured training data for our simple linear regression model.
A scatterplot of height vs. annual income. Each blue dot is an individual observation.
The linear regression line (red) plotted against the true data (blue)

A common classification model: KNN

A hand-labelled scatterplot. This example is inspired by CS109A, a course offered at Harvard College in the Fall of 2016.

Lastly, a (very) brief discussion about artificial neural nets…

A toy neural net with 9 artificial neurons. Borrowed from https://en.wikipedia.org/wiki/Artificial_neural_network

Key Takeaways:

  1. Machine learning combines computer science and statistics to create statistical models, which are then used to make predictions about the world or to infer patterns in your data.
  2. Statistical models are really just mathematical functions (e.g. Y = m × X + b). They are determined by their parameters (e.g. m and b), which are learned via the training process. Models also have hyperparameters (e.g. the K in KNN), which are tuned by trying out many possibilities.
  3. ML models learn from training data, which captures our knowledge about the past.
  4. There are, in general, two types of supervised models: those used for regression (like simple linear regression) and those used for classification (like KNN).
  5. Regardless of whether or not you’re building these models yourself, there are a handful of important ethical questions that require deep thought before you take action on the results of an ML model.

A parting message:

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store