for dataset

[DAOUDISARA2]

Is there a way to do a k-means clustering on data (file.txt )? Please, help me...

[LutzCle]

Hi,

I assume you're trying to load a plain text file? Clustering requires numeric data, e.g., a matrix of floating-point numbers.
As this is a research prototype, CL-kmeans only supports a very simple binary format. I guess it wouldn't be too hard to write, e.g., a CSV-to-binary converter. If you're interested in contributing a converter, take a look at binary_format.cpp and file_info.py.

[DAOUDISARA2]

hi,
I use the data set from benchmark Rodinia : https://github.com/JuliaParallel/rodinia/tree/master/data/kmeans

[LutzCle]

Ok, in that case there are two options. You can either convert the file, or tweak the generator to generate the binary format. You can take a look at my data generator for an example of how to generate the binary format.

[DAOUDISARA2]

Thank you, in your article "Efficient and Scalable k-Means on GPUs ", how you applied your binary dataset with kmeans Rodinia benchmark, please I need your help because I applied your binary dataset with kmeans Rodinia and it does not work.