WhY DoEs ThE SaMe line fit all these datasets? Hey, If you are ‘data-driven’ then decision-making can seem deceptively easy. Ever compared a bunch of descriptive stats between datasets? Like mean, standard deviation, correlation? It is easy to look at a table of such numbers and carry on with your day. Sadly, using popular stats… Continue reading Anscombe’s quartet and statistical noob traps
Category: Main
Understand matrix decomposition to understand LoRA and QLoRA
Intuition and Linear Algebra behind LoRA and QLoRA.
What is a standard error?
Hi, Most people understand standard deviation. If not they at least understand how to derive an average. I have had a few people ask me the definition of standard error. So I suppose there is an impetus to make this post. Posting this on the 1st of April isn’t a jab at the people with… Continue reading What is a standard error?
Days of Christmath
Merry! I bring presents regardless of whether you celebrate the birth of Isaac Newton… according to the Julian calendar. I know everyone is excited to peruse math over the holiday period. Anyway, on the first day of Christmath your good friend brought to you a song and a math. December 25th: Day 1 Area of… Continue reading Days of Christmath
Fashion MNIST and why it’s better than MNIST for modern Computer Vision
Hey. Many people are familiar with MNIST, if you aren’t then it is a dataset consisting of handwritten digits from 0 to 9. It stands for Modified National Institute of Standards and Technology. It’s frequently used in training computer vision models to benchmark various classification algorithms. Grad projects and research papers alike use it to… Continue reading Fashion MNIST and why it’s better than MNIST for modern Computer Vision
Practical SQL for everybody
Hay Everyone’s favourite querying language can do more than fetch data to analyse in Python. I firmly believe that in the darkest timeline people use SQL to do everything from developing apps to creating mission critical artificial intelligence. SQL is easy to pick up and do cool things with, which makes it a good stepping… Continue reading Practical SQL for everybody
Unsupervised machine learning to cluster user data
Ohaio, It is interesting to analyse segments created by algorithms. Product managers and marketers will find it a somewhat unbiased way to make sense of customer data. The algorithms can work well but it helps to have a sufficient grasp of statistics and understand the limitations of each model. I thought it would be useful… Continue reading Unsupervised machine learning to cluster user data
Robust Principal Component Analysis for everyone
Bonjour Most people have heard of Principal Component Analysis. It was the last thing I learned and perhaps many people learn in their introductory linear algebra course at university. I’ve had to use it plenty of times to reduce dimensions when staring at a dataset with too many columns. Many people are satisfied with vanilla… Continue reading Robust Principal Component Analysis for everyone
Decomposing ARKK portfolio returns into factors
Hi, Some people like to watch the world burn. I’m not one of those people. I like to watch it dismantle. That is why we’ll be linear dismantling returns from an expensive ETF or maybe a mutual fund. If a person invests in mutual funds or pays a fund manager a fee, it is the… Continue reading Decomposing ARKK portfolio returns into factors
Accounting ratios for apes
Namaste, Accounting is the process of identifying, measuring, recording and communicating financial information. If you are not yet asleep then please congratulate yourself. Accountants are reputably boring but they are sticklers for keeping records of a company’s assets, liabilities and owners equity. This is useful for people in finance that want to compare similar companies.… Continue reading Accounting ratios for apes