A FastAPI service that scrapes SEC Form 4 insider trading data from openinsider.com, stores it in SQLite, and serves it through a secured, rate-limited REST API — built as a backend data layer for my trade journal.

I love streaming algorithms - they make you think! Since they can process data in (near) real-time, they are ideal for applications where immediate feedback is needed. Most of today's recommendation or machine learning systems are built on batch process, however.

As I was contemplating on how to best display the dashboard for stock analysis, I came across this Nuts and Bolts of Chart Types - I like its sarcastic tone, and to be fair, some of them have truth to it. I am going to leave it here for my future reference.

While their transformed data may not be used for further data analysis, T-SNE and UMAP (non-linear) dimensionality reduction techniques provide great values for data visualization!

Values of the features or variables in the dataset are not the same. Even if they are numerical values, their ranges can be quite dramatic. Ensuring that all features are treated equally in terms of scale and range is essential for the performance and stability of many machine learning algorithms.