The Personal Analytics of My Life (2012) | Hacker News Reader

The Personal Analytics of My Life (2012)

wired.com

100 pointsarchagon21 comments

I have been doing "self-tracking without tools" for about two years now. I described my approach in this essay last year:

https://www.gibney.org/a_syntax_for_self-tracking

I can not see me ever going back to not doing this. It is just too interesting to be able to look up and analyze all the data of what I did and how I felt.

Additionally, it improved my ability to notice and quantitatively express what I feel.

My expectation is that in the future, this type of self-tracking will become very common. Not only for the sake of curiosity. I expect that it will also turn out useful to do "randomized controlled trials of one" which will result in meaningful results for the individual on how their behaviour impacts their mood and health.

Does "analyze" in your case mean skimming through past events or something more involved, as in the Wolfram's article? I've been logging my video game and work sessions in a similar manner for ten years and always wanted to add a statistics report to the custom tool I use, but my ideas stop at simple stuff like longest N per period or bar chart distribution per period.

I started writing an analytics pipeline for this type of log about a year ago and worked on it on and off. Interestingly, it turned into something close to a jupyter notebook alternative that runs in the browser.

The part that is most tricky is to do treatment-effect analysis with this type of irregular data. I have written a module for it and it looks promising. For example: When first analyzed a ton of treatment-effect-pairs, the one that strongly stuck out is that Ibuprofen causes a drop in headaches :) A sign that the approach works. Among the promising relationships I see is that it looks like Vitamin D3 intake reduces symptoms of the common cold. I have found some of these "promising" pairs, but I need more data before I consider them relevant enough to publish them.

Very cool.

I feel like the pipeline of knowledge could do with more published lists of "promising pairs" that others could follow up on.

It will be an interesting data world when everything significant we do is auto logged, hopefully in personal private logs, by our smart watches or glasses or something. Seems inevitable.

Then we will finally find out if an apple a day really keeps the doctor away!

You might want to try org-mode

Stephen Wolfram’s personal analytics (also described here - https://writings.stephenwolfram.com/2019/02/seeking-the-prod...) - have been the inspiration for the “dogsheep” project from @simonw. (Wolf… ram… … dog… sheep, get it?)

See for example: https://datasette.substack.com/p/dogsheep-personal-analytics...

My experience with dogsheep (healthkit-to-sqlite in particular) is there are multiple (two, in my case) show-stopping issues around usage, that have open pull requests with fixes the author is not merging despite doing occasional releases.

I ended up using it briefly but hand patched the python PIP installed to get it running. It's difficult to recommend after that.

This month I've started journaling every day. I'm looking forward to doing some analysis on the journal once I have enough data. I'm intending to do sentiment analysis to plot my mood over time and try to correlate that to words written and daily goals accomplished.

One thing I didn't really get from Wolfram's analysis is how this data drives decisions. It's neat to look at, I read the section on email and skimmed the rest, but I don't see how consequential it is. I hope to use my analysis to decide whether journaling is improving my mood and whether I should focus more or less on goals. In November I'm also going to start marking approximate exercise time and grade myself 1-5 on diet. If sentiment analysis plots work, that can help me dial in exercise and eating habits. That's the idea any way.

I'm tempted to wear a fitness tracker, that would give more data and could automate data collection for steps, exercise, and sleep - but I find the wrist trackers uncomfortable to wear.

Planning to do sentiment analysis in the future on your own writing that you’re also planning to do in the future seems like using a cannon to kill a mosquito—way overA-powered and unlikely to yield nuanced results anyhow. Why not just add to your journal entries a few quantitative (incl. multiple choice) measures of how you felt each day? Then there’s need to deduce anything about how you probably felt—it’s right there in black and white, and with much more nuance.

The rest of your self-tracking sounds like a fun project.

This is basically the Quantify Self movement. I have a picture of every meal since 2005. Currently past 15K images. All taken under similar lighting and angle in the hopes I can analyze later. Not knowing ML was coming. i may still do that but what I wish I did more/better job of is health. If in a spreadsheet let's say noted things like mood, energy, times of most production, life happiness etc. Now in my 40's with aliments. I wish I had the data to see maybe when/where/what about things that are taking place now. Including eccky things like BM info like when, type etc... also including lipid panel info etc when you typically do a full blood workup. Moving job to job diff health care to health care. There is no exportable consistent dataset you can "carry" with you maybe until now but even now it's painful as they don't allow for easy export.

How many times have you gone to your doc and get asked a basic "when did this start?" question and/or what were you doing 2 mons when it started.

There is a evolution in health care monitoring going on which can lead to incredible insights into your health both mental and physical, but then so does the world where it may judge you based on what you have now that you can't keep anything a secret.

Wolfram also streams on twitch and sometimes has open design review sessions of new features in his software

Discussion from 2012:

https://news.ycombinator.com/item?id=3680283

I was hoping for a list of items that this person tracks. For example, I only manage to track my lifts and run times on a daily basis. I would like to measure blood pressure, sleep time, amount of sugar consumed, eyesight, and working memory, and maybe typing speed too. What about you, what's important to you?

Number of emails is so, so far down my list of things I care about.

Weekly body weight in Excel, and then I have a photo album for corresponding pictures on my phone.

Electricity usage month to month which I’m hoping to correspond with new data: set up an arduino to determine average indoor temperature in my house, and then also find a weather api to get average outdoor temperature in my city.

oh man Stephen wolfram being an inspiration again

Cool stuff.

When I started reading this I thought it looked familiar. I remembered reading this years, ago, and sure enough the article is from 2012, so needs a (2012) after the title.

Reply on news.ycombinator.com