Define the benchmark for spectral clustering 2.

[ypriverol]

We have to design the spectral clustering benchmark, the following preconditions are needed for the test:

• A single machine, with local space for stable tests.
• Different sizes of datasets.
• Set of well-defined metrics for performance (time, cpu/memory consumption, ) and clustering quality.

[ypriverol]

I have to check my local Machine and we have:

Memory: 15.6GB
Processor: Core i7 CPU 870 2.93GHz x 8
Disk: 500G

@jgriss let me know (with a +1) if you think this is fine and I tick the first precondition in the issue.

[ypriverol]

We should define datasets of the following size:

Performance datasets: 100k, 1M, 2M, 5M, 10M, 100M.
Quality Datasets: Synthetic Peptide dataset.

[jgriss]

Suggestions:

• Melanoma dataset
• Kuester peptides
• PRIDE Human
• Gigy
• TCGA Colon Cancer

[ypriverol]

Create a user for @jgriss in the machine for testing.