Young Hands Club

July 28, 2019

Week 2 : Progress Summary

Filed under: Shreyas Ragavan — Tags: — Shreyas Ragavan @ 4:20 a.m.

I've attained a better idea about the WoT (pending comments from diana_coman). It was interesting to compare it to the only 'social' network I've been regulary active on : Linked in.

I implemented HTTPS for the website, and enabled pretty links to the posts and pages on the blog. However, I am yet to attain clarity on why HTTPS is needed for my website / blog, and diana_coman's comment that 'HTTPS don't work', has been puzzling me. I'm yet to make time for this, but as a start I've gone through Let's Encrypt's Subscriber agreement and do not feel encouraged. My notes on this are incomplete and thus not posted.

This week, a lot of time was spent in studying the UMAP algorithm. This is basically a dimension reduction algorithmn, and is relatively new. The concepts are very mathematically involved, and there are not many instances of online documentation in using the library. Though I'm not working out the original 51 page technical paper's math by hand, I've made strong strides in gaining an intuition of what the algorithm does. It seems clear that the hyperparameters of the algorithm, (especially the distance metric) have to be tuned for the data, on a case to case basis. It is not yet clear whether it makes a difference to the machine learning model, whichs is the subsequent step. This would be my first major exploration in data science at the algorithm level.

This kind of progress, and gaining such clarity has been rare for me and I crave for it to happen all the time. My notes on UMAP are a work in progress, but I think it will serve as an excellent reference when completed, hopefully in the next few days.

I was also able to contribute usefully to some discussions on the Business Science University forum, and published my note about setting titles, subtitles and captions in a plotly object.

Some time was also thrown at tinkering with the aesthetics of my blog/website and I finally found something satisfactory. For the next week, my plan is to focus on asciilifeforms's v.py, and also get going with running eulora while trying to complete the UMAP project. I know Diana would say that's too many #1 priorities, but it seems like if I ignore the other projects - they will die. As such, I've already made some headway in my reading about V, and the EUcrypt series.

1 Comment »

  1. Not bad for week 2, eh?

    The plan for the week ahead should be a bit more fleshed out so that you can get to practice and improve also the tricky part of defining chunks of work and estimating time required.

    One significant reason for planning is precisely so that there is always space for more than just the 1st priority and nothing gets lost /dies (though at times the lower priority stuff may get postponed but not in the idiotic style of forever postponed and just carried around).

    Comment by Diana Coman — July 28, 2019 @ 8:11 a.m.

RSS feed for comments on this post. TrackBack URL

Leave a comment

Work on what matters, so you matter too.