no

Anyone here with good data mining skills?

I've been trying to research "gold standard" and "ground truthing" and have the feeling I'm missing some easy to find references. Is there some sort of "techniques for developing a gold standard" book/paper/chapter in a book that has a good summary? I've glued together some common approaches from a couple of papers but have yet to find an integrated document. I'm specifically interested in defining a gold standard for stuff that requires "expert knowledge" and/or crowd sourcing that is...given a bunch of sentences, are these sentences about X or Y or neither.
[grinding through sentiment analysis literature right now since that seems where this would be useful]

That's certainly how I got into programming but not really how I progressed. Hopefully that makes sense in a second.

I guess I got into poker most recently in my early poker days, writing ahk scripts to make for easier multitabling. After black Friday I ended up with a call center sales type job. The contact management system was so crappy that it took several clicks to do virtually anything and it wasn't integrated with the desktop phone applet. So I wrote some ahk scripts to do that stuff and found it really interesting so I decided to start learning programming for real.

At that point I mostly got enjoyment out of solving coderbyte stuff and completing online coursework.

Since then I've had the occasional idea to work on in my own time but the ideas are few and far between

Rusty I'd love some for instances of stuff you'd like computers to do for you that you haven't had time to tackle. I think maybe I'm just inclined to accept whatever imperfect but good enough solution already exists for most things

OK, well, here's some stuff off the top of my head.

I have a few hobbyist type CNC machines, and I write a lot of my own code for them. There are CAD/CAM packages but using them is usually seriously the long way around for things.

One of the things you find yourself having to do a lot is take an existing "drawing" (say, a collection of lines and arcs), and calculate an offset from your drawing. Do you see what I mean? Say you're cutting a 2" square with a 1/4" router bit. If you cut from -2, 2, to 2, 2 and then to 2, -2 and so forth, your square will be too small by the radius of the router bit. So you have to offset "outward" by 1/8"

I use a common library for doing geometry type stuff in python (shapely, which is based off libgeos). It has a function for doing this, and it even includes rounding corners where appropriate. "Great!", I thought

Except... it has a bunch of weird oddities. Recently I've been having problems where it will *reverse the direction* of my polygons, which makes all offseting operations backwards after that. And sometimes it doesn't close them properly. And sometimes it will change the "start" point of the polygon, which can be a problem if I'm counting on it starting from a specific place, which I often am. And, it makes curves by interpolation, which is both not great for CNC, and wasteful of time/memory/space/whatever.

I'd like a solution that works entirely on lines and arcs. I want to be able to scale and offset from a (probably closed) set of lines and arcs. I have something that sort of almost but not quite works.

And I don't really have time to fix it, because the shapely thing *mostly* works if I don't poke it too hard. But I can't give it away or use it blindly, because of it's oddities.

That turned out way more long winded that I expected. I actually have a lot of projects in the CNC space that need doing. I have a CNC simulator that I wrote, which mostly works fine but could really use some improvement.

I write a lot of software related to athletic training. Most of it is highly specific to me, though, I'm not sure anyone else would find much interest in it. Mostly I use them to plan my upcoming weeks, and evaluate performance over the training season.

I wrote some code to take timelapses with Canon cameras. It works pretty good but I need something to process the RAW files and produce more linear changes in exposure.

Looking at my projects those are the only things I'd kind of like to have working, that I don't currently have time for.

cool, what do you make with your CNC machines?

Do you know any wood or metal workers? Mostly, I make stuff to help me make stuff with CNC machines

I do some light robotics type stuff. A current project is sort of a "turret" that tracks targets with a camera. I'll also probably use it for algorithmic photo taking, something like a "gigapan" if you know what that is, and/or for adding panning to timelapses. I've made a lot of other random stuff for photography, basically various mounts and stands and stuff.

I've made a lot of decorative stuff. Something I've been waffling on lately is that I have a long bench, maybe 12' long and 4' tall, that is full of drawers. I haven't made drawer fronts yet, because what I'd like to do is make the whole set of drawer fronts into a mural. I'm stuck in the design phase, because I don't know what I want.

I make a lot of wood and metal working tooling. Clamps and hold downs and fixtures and stuff.

I dunno. I'm not really that productive with it. Everything takes forever, and I haven't really settled on anything that I'd like to make that people would want to buy. So you spend an hour of design/programming time for every minute or two of working time, you know?

I know "omg another framework!" but cycle.js is kind of blowing my mind right now. Anyone else play around with it yet?

When I was doing fire stuff, I knew one person who would make the tools for many people in our group. He'd also sell them at Burning Man. For stuff that you are going to char, he did some beautiful, though utilitarian work.

You live in Austin. It's only rumor, but I hear there are a few musicians in this neck of the woods. Have you considered building instruments or building for those kinds of people? I'd love to have a nice guitar stand that doesn't tip over.

I'd love to make guitars. I've gotten halfway into designing them a few times. Really I don't have the time for a lot of the stuff I'd like to do.

I'd actually really like to find something to make and sell. I don't know that I really even want the money that bad, but I'd like to make something unique that interests people. I am half putting together a website of various stuff I've made, to see if they are things people are interested in.

I guess this is why people do kickstarters. More than anything, to get something produced, I'd need time off and for that I'd need money.

I (somewhat) knew this guy back in LA.

http://www.woodbytoth.com/

The orb and vase are as many pieces of wood as it looks.

Very nice stuff. I can definitely say that I am not really that creative of a person and by far the hardest part of what I do is figuring out who to steal ideas from.

If you want to geek on guitar making, this video is pretty amazing:
https://www.youtube.com/watch?v=6vRcHEIawP8

and also this one, a bit shorter:

only a tiny bit, but it seems very nice from what i've seen.

plus side of weekly panic attack: finally got so fed up with having borked my .rvm permissions at some point that I finally fixed that **** and can run bundle install to my heart's content

Anyone working on multi-platform (win and osx) desktop apps? I have a hobby project I'm picking up after a few years. It's similar to football manager game. Is C++ still the way to go? I'd like to stick with open source tools and avoid a multi-language (objective-c + c#) solution if possible.

I'm not working on one, but I stumbled on Electron this morning and it seems pretty damn cool.

Yeah, I have multiple projects that are for OSX/Windows/Linux. A few even have android versions. I use C++, and my makefiles are works of art/garbage.

If it's going to have a GUI component, you need to pick how you want to do that. I like Qt for everything except how you build it (they want you to use their build tool, and not using it is kind of a pain)

So mostly if I have something that really needs an interface, I tend to make the interface in python, and supply the functions it needs via boost's python interface. I have also experimented with embedding Tk interfaces directly in C++ programs, which works... OK but not great.

Ha, so you've experienced the joy that is the Android NDK.

Interesting. I've used atom before and it is a native experience but had some performance issues. I don't have the skills to manage large amounts of business logic in JS though

I will be building a custom gui.How does Qt compare to the python approach? Is your main executable python that statically links all the business logic?

Even better, I have experienced the joy that is Java's C-to-java translation layer. Uh, what's it called? JNI. My god.

I typically write my applications as if the interface will be a separate component. That way I can change my mind about the GUI, use it with android, or even have just a CLI program.

So when I use python for the interface, I just make a python library that essentially provides a python layer interface to my C++ objects. So I can pretend all the C++ stuff is python stuff and proceed accordingly. I tend to use the python TK gui toolkit because it's easily installed everywhere, but Qt looks nicer imo.

You can use Qt directly with C++, it's fairly simple, but you'll have to use their tools to build your code. I can't think of what their tool is called, qmake or qtmake something. You also have to download and install their gigantic toolkit/library stuff. I just don't have a solid use case for it most of the time.

I'm not statically linking anything typically, except on Windows where I'll statically link to whatever libraries I'm using. If I'm using a python interface, though, the library for my code that I load in python will be dynamically linked. I'm not, like, distributing a re-compiled version of python or anything.

There's also pyQt

QtCreator

Is the tool Rusty is referring to. I've used it some. The thing I didn't like about QT was all the qt specific types required to interface to their API. It seems like a decent enough tool though but admittedly haven't used it that much.

No, I don't think I'm talking about QT Creator. I'm talking about "qmake"

http://doc.qt.io/qt-4.8/qmake-manual.html