Hi, maybe a silly question, but I often find understanding the variables helps me hold the process in my head. Why is “m” the variable for the size/number of instances in the training set, rather than “n” or “N”? It doesn’t really make sense to me.

Cut to the chase, I don’t think there is any compelling reason for it. The founding fathers in AI started out with certain conventions and some of them might have even been arbitrary. And as each of these conventions seeped into everyday use, we just let it be.

That makes total sense, thank you! Now is my obvious follow up: Why “J” for the squared error cost function? I assumed it would be “F” for function (or “G” for the next letter), or “C” for cost. Using “J” kind of threw me.

Hey @scoofy , in short, unfortunately, I don’t know. But out of curiosity, I googled a bit and find there are some discussions about your question (many people care about this). The most relevant one is relating J to Jacobi who was a German mathematician with some of his work related to optimizatio…

Thanks for taking the time, i certainly don’t plan on being this pedantic throughout the course.

It’s fine @scoofy , I am happy to help you learn better :slight_smile: Let us know when you have a question about the courses, and Cheers!

Why is "m" the variable for the size of the training set?

Course Q&A Machine Learning Specialization Supervised ML: Regression and Classification

rmwkwok June 20, 2022, 12:22am 4

Hello @scoofy, this is my guess.

When I learned Matrix in high school, we always said the size of a matrix is m x n, which is m rows and n columns.

In ML, we always represent a tabulated dataset as a matrix, having one row for one data sample, and one column for one feature.

So if there are m rows, there are m samples, and when there are n columns, there are n features.

You can also see m and n being used in wiki to talk about the size of matrix!

Hope this help!

Cheers!

2 Likes

Letters used to denote examples and features - are these standard across the industry and academia?

Topic		Replies	Views
Are x and m always the same? Supervised ML: Regression and Classification week-module-1	2	535	June 26, 2022
Parameters of the cost function Supervised ML: Regression and Classification week-module-1	1	488	October 2, 2022
Name 'm' is not defined in the cost function on the week 2 practice lab Supervised ML: Regression and Classification week-module-2	12	239	April 2, 2024
Letters used to denote examples and features - are these standard across the industry and academia? Supervised ML: Regression and Classification week-module-1	3	571	August 12, 2022
Some question about the element forming Cost Function Supervised ML: Regression and Classification week-module-1	2	509	August 7, 2022

Why is "m" the variable for the size of the training set?

Related topics