data-analysis

This is a general purpose collection of data analysis libraries for the Dylan programming language.

It currently consists of a statistics library providing some general statistical functions for calculating means, variance, deviation, and extrema.

Documentation for this set of libraries can be found online at http://opendylan.org/documentation/statistics/.

Contributing

Additional functionality and libraries are welcome.

Some immediate ideas for additions include:

• Mode
• Median
• Weighted means
• Trimmed means
• Skewness, kurtosis
• Absolute deviation
• Covariance
• Percentiles and quantiles
• Rankings
• Distances (probably a separate library within this repo)
• Z scores
• Distributions
• Random data generation
• Online / streaming descriptive statistics

Larger projects might include:

• Time series data support
• Dataframes
• CSV reader / writer
• Improving testing
• Benchmarks and compiler improvements (mainly with LLVM)

Building

This library can be built with Open Dylan 2014.1 or later on these platforms with the C compiler back-end:

• Mac OS X
• Linux, 64 bit x86_64
• FreeBSD, 64 bit x86_64 (not available with 2014.1)

Using Open Dylan from current master with the LLVM compiler back-end should work on 32 bit and 64 bit Mac OS X, Linux and FreeBSD.

The HARP compiler back-end has a bug that prevents it from being able to compile this library.

To use this library in the typical manner given Dylan's lack of a package manager:

• Use a git submodule to pull this repository into your project.
• Add registry entries for the libraries from this repository that you plan to use.
• Use it.

Documentation

Documentation is built using Sphinx. This can be installed via:

pip install -U Sphinx

Then:

cd documentation
export PYTHONPATH=/path/to/opendylan/documentation/sphinx-extension
make html

To build the documentation, you will currently need a clone of Open Dylan from git to have the correct version of the Sphinx extension. (Or you could clone that locally and use the path to it directly.)