Hacker News new | ask | show | jobs
by pat2man 4614 days ago
I've been thinking about this for a while. It seems almost impossibly easy and yet no one has really done it.

Maybe it's because the consumer startup crowd and the data science crowd don't overlap enough. If http://prediction.io supported time series data you might see a few of these products pop up overnight.
2 comments
driverdan 4614 days ago
> It seems almost impossibly easy and yet no one has really done it.

How does it seem easy? If you just want to track one or two data points and put them on a graph sure, it's easy. If you want to track 100 sources of data, each in a different format with different types of data it becomes quite hard. Then try taking that data and analyzing it. Try looking for correlation between your manually recorded caffeine consumption and your sleep patterns logged with FitBit / BodyMedia / another service.

It's a hard problem with a limited market and limited options for monitization.

link
manjitjohal 4613 days ago
>It's a hard problem with a limited market and limited options for monitization.

I disagree; I think there is enormous potential in this market; having a set of concise and insightful information that shows you are making progress with a particular goal is something a large number of people would be interested in.

link
ivankirigin 4614 days ago
I never said it was easy. I completely disagree with that.

Machine learning, simple design, scaling 100 API integrations, and mobile apps are all really, really hard.

link