we are running the kmeans algorithm both in C++ (our own serial and parallel implementation) as well as in Python using the serial implementation of Scikit-Learn. The aim of the project is to demonstrate the performance increase when switching from serial to parallel programming, plus we decided to spice the things up comparing how fast the serial C++ implementation is against the one in Python from a well-known library (so we assume it's a state of art implementation!).
Because the algorithm is intrinsically non-deterministic, we decided to start the algorithm with pre-defined datasets to decrease variance between experiments. Please check the folder input for more infos about the datasets we used.
We included an easy script (that is, prepare-project.sh) that helps configuring the environment on your virtual machine.
All you need to do is to install git with
sudo apt-get install git
and then clone this repo:
git clone https://github.com/A7F/aca-kmeans.git
move to the newly created folder with cd aca-kmeans and assign run permissions to the script prepare-project.sh with
sudo chmod 777 ./prepare-project.sh
Please remember to set the correct working directory inside the script you want to run first! For example, if you want to run
the serial script, paste your path inside the variable base_dir. After that, the bash script we provided automatically
builds the project and prepares the runnable file in the same directory, so all you need to do is to call ./aca_kmeans.
follows a small section of useful links and resources to keep in hand, you know... just in case...