Show HN: I wrote a book about using data science to solve “everyday” problems
andrewnc.github.io
andrewnc.github.io
So I wrote "Everyday Data Science" which is a collection of stories, tutorials, jokes, math, and code all written to inspire people to analyze their personal data.
In general, I was also inspired by the challenge to "make $100 online" which I have done in the past month since launching. It was daunting, and I felt quite vulnerable, but overall I'm pleased with what I've made.
I wrote up this quick post to give you an idea of the process I followed to write the book, and some of the content.
I'd love to know your thoughts and am open to (nice) feedback :)
I think books like these can be a great eye opener. We all remember thinking "what am I gonna use this for??" in high school maths, physics etc, and I think this is a fun, approachable and interesting way to see real life impact of maybe otherwise dry and abstract stuff.
You always did make me thirst for more understanding of ML... guess I'll have to buy this book. Do you make more margins on digital or print copies?
By the way -- HNers -- if your company needs talent in data science, Andrew is easily one of your best candidates. His intuitive understanding and teaching of data and ML inspired me to be a better scientist. Andrew thinks critically and is also a great person to work with.
Those are kind words coming from maybe the single greatest programmer I know.
Thanks for your support and for the recommendation :D
The unfortunate truth is that book writing is a low margin game and you shouldn't do it for the money. Development or actual data science is far more profitable.
But there are many other intangibles :)
This book resonates with me.
This reminds me of the "data science" book was from Data Smart, from the Mailchimp CDO, talking how to keep orange juice tasting the same all year round (using a seasonal fruit) or calculating the likelihood of a consumer being pregnant - all within Excel.
Motivation is such a fickle beast. For whatever reason, I felt good during the entire writing process. Going back and forth with the editor was definitely more challenging.
One thing that helped, actually, was I tweeted an artificial deadline when I started writing. That was immensely motivating for me. I ended up missing the deadline (again because of editing woes) but that was key in helping me push this over the finish line.
Again, I'm glad this was motivating for you and I wish you the best. Feel free to message me on twitter if you ever wanted to chat more about projects you're working on.
So I guess that's a long way of saying no, there isn't another option for now.
EDIT: Nevermind its in the post and I was confused: https://gumroad.com/l/everydaydata
The book looks very intriguing! Before buying it, I wish I could read a bit more about what one could expect from the book. Let me explain :)
> This book is for people as untechnical as my Mom or as technical as my Applied Scientist friends working in big tech.
> ODEs On A Diet
Do you present an intuitive explanation of ODEs? Would both your mom and an Applied Scientist find it informative? If yes, I would love to read! A few pages from this chapter would really help in making a purchasing decision.
> a solution to Multi-Armed Bandit problems which are significantly more efficient and only takes a few lines of extra code.
Does the book include code samples? As someone wanting to learn about, e.g., ODEs through code, I would love it code samples were present in the book.
I understand that $7.99 is not a lot in the US, but it is not the case in a few other parts of the world. Given that there are no refunds, having more information about the book would be valuable for international customers.
BUT... the books looks great, and big congratulations on publishing it!
I've submitting a petition to "claim" the book as an author.
In my experience, the most difficult parts of the process are (1) translating from qualitative problems “in the field” to a formalized technical problem, and (2) all the wrangling necessary to implement the formal problem using field-collected data.
I find that I spend most of my time as a data scientist working on these parts of the process, and often don’t have the bandwidth or even requirement for more advanced methods. Not that fancy techniques are the goal, per se, but I do notice I rarely have the opportunity to use them.
It reminds me of the problem of a car noise, but you bring it to a mechanic and it doesn't reproduce on the spot so diagnosis is extremely hard. In fact I've had a heart issue of this sort recently where I experience symptom intermittently, sometimes weeks or months apart. My cardiologist has brought me in multiple times for an EKG, but since itt not while I'm having the symptoms we have no idea if the normal EKG readings are telling an accurate story.
Essentially I want to see personal health devices like the black-box on an airplane. Take blood pressure, weight, and resting heart rate once a week? It goes to the black box. Log a physical symptom or mood? Goes to black box. Sleep and activity patterns, I integrated smart-watch EKG? Goes to black box.
I think the advent of truly seemless UI and UX with unobtrusive comfortable devices that could do this would provide a massive leap forward in preventative health.
We just need to get there in a way that doesn't make it one of the most massive land-grabs ever of personal data by device vendors.
It's similar to what you're describing and aligns well with the idea of using your personal data in a more open way.
I tried for a while to get the epub / mobi version but the PDF was always mangled by every conversion technique I tried.
I did just get a nice tip from a HN user about a piece of software that might work. I'm going to give it a try and maybe I'll have an epub one down the road.
But as of now, there isn't an epub version.
I've been working on converting an old AI book into cleaner markdown for a while, and the biggest issue there has been the lack of a clean source. I'd be surprised if there isn't a way to manage this conversion.
It's not as nice looking (personal opinion) but it is functional at least.
I always get annoyed when I navigate to a page and there are HUGE buttons telling me to spend money. So I tried to not do that.
Although, I may have pushed it too far in the opposite direction.
Anything plans that address this?
I settled with the PDF as a soft copy instead (and I put PDFs on my kindle, even though that's obviously not desirable).
If there is a nice way to get PDFs to read natively on kindle, I would jump on that, but I wasn't able to get it to work.
Book looks interesting, btw!
Yes, in theory.. but in practice it didn't work when I tried that route. Because I use the Tufte-Book template, there is a constant margin to the side of the main content that holds figures, equations, and such.
This margin gets brutalized by every program (even the propriety ones Amazon built).
Lessons learned. I need to be more mindful of the various conversion processes in the future.
This?
https://github.com/Tufte-LaTeX/tufte-latex
I've often been disappointed at the type of html it's possible to force out of (La)TeX. And epubs are pretty much html, css and images.
I generally think that some kind of markdown with the help of pandoc is the happy path for pleasant writing/editing and good output for html/epub and print/pdf.
I did find this, that have some hints on how to get xml, and then xhtml with support for equations - but it looks cumbersome:
https://tex.stackexchange.com/questions/1551/use-latex-to-pr...
See also: https://github.com/duzyn/tufte-markdown And in particular (beautiful!): https://edwardtufte.github.io/tufte-css/
Did you cconsider selling access to the source (eg: private github repo, suitable license)?
One option for getting help would be to open up everything to everyone - selling access to the source might be a way to retain some income while enlisting some help in conversion from volunteers (yes, we would first pay full price for the content, then volunteer to try conversion options to produce passable epub output).
The only alternative I can think of is copy the paragraphs to html p tags and then any equations written in latex could be converted to images with LatexIt! (if that’s still a thing)
That all assume it’s enough if a deal for you, understand if it’s not and thanks for the reply
I'll update here and on twitter when it's ready.
I also sell the epub directly on gumroad bundled with the PDF https://gum.co/everydaydata
https://andrewnc.github.io/preview/page_20.png https://andrewnc.github.io/preview/page_94.png
That should give you an idea of the style of writing etc :)
(That being said, I love the idea, and hand-drawn visuals + colorized equations!)
Thanks for the constructive feedback.
I'll update here and on twitter when it's ready.
My first two immediate impressions: 1) The handdrawn illustration thing is cutesy and cloy. 2) I've no idea why the line spacing is so massive. The word density per page feels like half a normal book.
Of course those are just superficial remarks, and I have barely read anything yet, but they both immediately annoyed me.
*the line spacing thing is annoying because it makes me feel like i'm constantly spinning the scroll wheel on my mouse more than actually reading the text.
As far as the drawings, I'm not sure what you mean by cloy?
As far as the handdrawn graphics, I think it's kind of a schtick that XKCD created and it works for them, but I find it really tedious when others imitate it. The example chart, something about birds drawn on top of derivative symbols, didn't make much (any?) sense to me.
I appreciate that the book has brevity, and I want to motor through the text. I guess if typesetting is the main problem, that's a good thing. Probably among the easiest things to change.
I sent out an updated version on gumroad with new line spacing, I hope that is easier to read :)
If sold out, will it be available again soon?
Edit: refreshed and it's there again!
I find that a quite strange way of thinking...
I don't judge a restaurant's "value for money" based on their dollar per calorie. I don't choose a laptop based on how many components it has. I don't choose to watch a movie based on it's bitrate.
I kinda get that page count is an "easy metric", but iy seems super shortsighted to chose to not buy a book based purely on page count.
And even at $18, that's, what, a movie with a candy bar trip? A sandwich and a couple of coffees at a cafe? Cover charge to see a band at a bar? There seems very little chance that most people who read this site wouldn't get a couple of hours entertainment from a book like this, with at least as high a likelihood as enjoying a movie/band/cafe recommended here.
Having said that, I'll buy the PDF too, but only because it's my second preference after a Kindle Edition, which doesn't seem to be available...
I also agree that Kindle would have been awesome :/
It's the best metric currently available for a book that has no reviews, and is from an author I haven't read anything else from.
You have a minor typo on the page - 'enthrawled' should be 'enthralled'
I hope your niece likes it, I would honestly love to know how she handles it. I bet she'll get a lot of the ideas, and might have to skip over some of the nitty-gritty details.
Sure that it's not "entrawled in data", like someone entangled in a swift web of information? ;)
She's a brilliant woman whom I respect very much. I didn't intend for my phrasing to be denigrating and I apologize for how it is phrased.
Really nice work, I just bought it and put in on my reading list for today
If you want the book reading for errors, I would be happy to do so.
Can you update the book on the fly?
Offer stands. If you want it proofreading, I will do it for a credit in the intro page or similar.
Page 17 of the book:-
Fit: The process of showing data to a model to the model get better at explaining or predicting something.
seems like a typo