Nice writeup. Did you consider using Google's Prediction API before rolling your own recommendation system?
Embarrassing Flaws That Made My ‘Weekend’ App Possible
51–60 of 69 posts
Re: Embarrassing Flaws That Made My ‘Weekend’ App Possible
#52Earlier quoted context omitted.
Personally, I keep a timesheet on when I started and finished working on something, for instance Tue 19/10/2010 | 15:40:00 | 17:10:06 | 1:30:06 | Looking at admin interface for sensor daetails. Trying to fix None issue, seems to be related to subclassed models. Adding code to twisted for CT settings Turns out that one of my Django models is breaking the Django admin interface, and also that I have to have calibration…
Do you find that it helps? The closest thing I got that looks like what you got above there are my git log messages!
Seconded. Using distributed RCS, you commit often and therefore have a pretty good log of your daily activities.
Re: Embarrassing Flaws That Made My ‘Weekend’ App Possible
#53Nice writeup. Did you consider using Google's Prediction API before rolling your own recommendation system? http://code.google.com/apis/predict/
Possibly ruled out because : Access to the Prediction API is currently by invitation only. To request access and learn about the access process, sign up on our waitlist.
http://code.google.com/apis/predict/docs/getting-started.htm...
Re: Embarrassing Flaws That Made My ‘Weekend’ App Possible
#54Another great story on 30 days to an "overnight success". Launching a product, even a "weekend project" is just damned hard to get right, end-to-end. Harder than you think out of the gate. I love how this puts things into perspective.
Actually I think this just raises the bar - "Look, it only took 2 hours * 2 weeks to make something useful."
Now everyone who has been slaving away for >24 months to make their own end-to-end power monitoring solution feels slightly lame.
Look ma - power monitoring in only 1200 hours! (http://gridspy.co.nz)
Re: Embarrassing Flaws That Made My ‘Weekend’ App Possible
#55moconnor: did you keep a journal while you were building this app? I'm thinking to myself as I am reading this that there's no way you could have gotten from A-Z, then only after that started filling in the blanks accurately about what you did on day #X. (Heck, I can't remember what I ate yesterday!) Motivational, good stuff, thanks for sharing. Now I'm thinking of an app for HN'ers who just need a little help gettin…
Great idea, please do this. It should also record the choices encountered at each step and tabulate the paths not taken, which would be very useful if something doesn't work and you need to backtrack.
Re: Embarrassing Flaws That Made My ‘Weekend’ App Possible
#56I haven't looked at this data to know if this would fit, but SVD worked really well for the Netflix prize. When I first heard its use in recommender systems I laughed and then thought... wow, that's kind of perfect. Here's a good explanation if you're not familliar with it, and very accessible to those who haven't studied linear algebra (or haven't thought about it in a long time) http://journal.batard.info/post/2009…
+1 to this... I've had very good luck using SVD/LSA-ish approaches for building recommendation engines. HOWEVER- the big catch with naïve (i.e., simple to implement and understand) approaches to LSA are that every time you add a new datum, you have to recompute the SVD of your data- which is time consuming, and raises some interesting logistical challenges when done in the context of a web app (i.e., how you persist…
NumPy is not available on the Google App Engine: http://code.google.com/p/googleappengine/issues/detail?id=19... Apparently they don't want to vet C libraries?
Some searches suggested: http://www.picloud.com/
Re: Embarrassing Flaws That Made My ‘Weekend’ App Possible
#57Earlier quoted context omitted.
Great, I'm glad to hear it's proving useful! I'll do a follow up with how well my NoDB implementation stood up to the pounding it's getting at the moment later in the week...
Here's a feature suggestion for you from a fellow BGGer: Allow users to restrict the search to games from with specific amounts of players.
Yeah, #players and playingtime are both on my list of things to optionally filter. Just need to find a nice way to do it without complicating the initial user experience.
Re: Embarrassing Flaws That Made My ‘Weekend’ App Possible
#58moconnor: did you keep a journal while you were building this app? I'm thinking to myself as I am reading this that there's no way you could have gotten from A-Z, then only after that started filling in the blanks accurately about what you did on day #X. (Heck, I can't remember what I ate yesterday!) Motivational, good stuff, thanks for sharing. Now I'm thinking of an app for HN'ers who just need a little help gettin…
I didn't keep a journal, but after a couple of weeks when I decided to write this post I jotted down what I'd done when. The filesystem helped with a lot of dates (creation, last modification and so on). Going to England in the middle was a good, specific memory aid too - it's easier to remember what I was doing and thinking in an unusual place.
I love your appjournal idea!
Re: Embarrassing Flaws That Made My ‘Weekend’ App Possible
#59Nice writeup. Did you consider using Google's Prediction API before rolling your own recommendation system? http://code.google.com/apis/predict/
No. I thought about Directed Edge, but playing with recommendation algorithms was the raison d'etre of the project. I didn't want to make a web site and throw some amazon referral links at it, I wanted to play data miner!
Re: Embarrassing Flaws That Made My ‘Weekend’ App Possible
#60Wow, great idea. I am envious.