Wiki has pictures.
- What a Qt application which displays the distribution of distances between random points inside a unit square.
- allows the generation of random point clusters - uniform distribution in a square or normal distribution with customizable stddev and clip radius
- allows the user to distort these point cluster by applying affine or projective transforn from the tip-of-the-mouse pointer
-
Why: machine learning usually involves clustering. If you know how many clusters to seek, the problem is simpler than when you don't.
I wanted a tool to explore what a computer would "see" if examining only the distribution of distances, perhaps such information may be used to in separatings clusters. That is to say, this is rather a toy for a human mind (mine) to explore a bit the problem -
How to build:
- Qt 5.7.0 - may work with earlier, I haven't tried
Just make sure you have the QChart module installed - [Eigen3.3](http://eigen.tuxfamily.org/index.php?title=Main_Page)+
- a C++ compiler able to do C++11 compilation which is also supported by Qt
- QtCreator 4.1 set to your liking (optional) - alternatively, just use `qmake` after adjusting the `distspectrum.pro` file to build your desired debug/release target (if you don't intend to develop, but just to use the toy, I strongly urge you to build the `release` version if you value you time).
- adjust the `EIGEN_DIR` value in the `distspectrum.pro` file
- Build the project your usual way of building a Qt application