Dec. 30, 2024 Supervised Learning Supervised learning is a fundamental type of machine learning where we learn from examples that have been labeled with their correct answers. The term “supervised” comes from the idea that some knowledgeable entity (the supervisor) has already gone through the data and provided the right answer for each example. This supervisor could be: Human experts manually…
Mar. 15, 2024 [This is a short story written using Claude 3. Illustration created by DALL-E using the story as the prompt.] The fluorescent light flickered overhead, casting a pallid glow on the fading posters tacked to the cubicle walls. Jyotirmoy leaned back in his chair, the old wheels giving a tired squeak of protest. His eyes drifted across the familiar images - Warhol’s lurid pop art, the…
Oct. 20, 2022 This is a review of Claudia Goldin’s ‘ Career & Family ’ (2021, PUP). At the outset, let me confess that I’m not very well-read in this field. So these are the general impressions of an economics student and parent and not an expert review. Goldin’s book is subtitled ‘Women’s century-long journey towards equality’. The women here are college-educated women in the US. There are…
Jan. 20, 2022 It was time for a new laptop, the screen had started to glitch, possibily too shook up from being used while travelling on Delhi roads. I got this machine at around 60% of MRP on Amazon . It is tested to military standards, so I’m hoping that it’ll last longer. [I know, I know, militaries are huge machines, and the standards tested against may be closer to those for toothpaste than…
Oct. 3, 2021 It all started off with a poor economist running this R code: N = 520000 ALPHA = 0.01 y = rbinom (N, 1 ,ALPHA) mod = lm (y ~ 1 ) print ( coefficients (mod)) The answer should be something close to ALPHA , i.e. 0.01 . What R was printing out was weird numbers like 1e-10 . This was the evening before an important presentation. The economist had written the code snippet above after…
Jun. 27, 2021 This little special relativity problem occured to me today. The setup As usual Einstein is conducting an experiment in a moving rail carriage. He is observed by Jyotirmoy who is sitting on the platform. The carriage is moving from left to right. Here is what the experiment looks like from Einsten’s point of view. He has a powerful computer with all kinds of fancy peripherals sitting…
Jul. 24, 2020 I completed by online measure theory lectures today. Thanks to all those who participated. Here’s an archive of the content. Lecture 1 Motivation: the strong law of large numbers. Preliminary notions from set theory: countable and uncountable sets, algebra of sets. Slides Lecture 2 Construction of measures: Caratheodory’s theorem. Application to the infinite coin toss example.…
Jun. 20, 2020 I’ve been thinking about how to teach measure theory starting from building a probability measure on the sequence of infinite tosses of a fair coin instead of the standard treatment which starts with Lebesgue measure. This post is an outline of how to translate Vitali’s proof of the existence of nonmeasurable sets for Lebesgue measures to this setup. The Problem Our sample space Ω…
Feb. 16, 2019 Economists discussing machine learning, such as Athey and Mullianathan and Spiess , make much of supposed difference that while most of machine learning work focuses on prediction, in economics it is causal inference rather than prediction which is more important. But what really is the fundamental difference between causal inference and prediction? Let us try to define both tasks.…
Dec. 1, 2018 I needed an offline copy of the Rust Book to read on my commute. The authors have been nice enough to share the source files for the book on GitHub so I could generate an EPUB with Pandoc. I have put up the scripts I used as a pull request on the Rust Book repository, but meanwhile you can download the book from here . Please write with your feedback to jyotirmoy@jyotirmoy.net .