Hands-on sessions - omniclustering
Resources
Choose a module to add to the Omniclustering benchmark. Alternatively, you can also work on the Iris example from the step-by-step guide.
Projects assignments
To avoid overlaps, please specify which projects you are working on:
https://docs.google.com/spreadsheets/d/1nnPSALl6up0-yjXiWLPe0R7gSUyveozUQdVmZLu2fno/edit?usp=sharing
Project creation
See the Create a module - Clustering Omnibenchmark links to prefilled projects.
Warning
Please use an explicit name for easier recognition and also to avoid names overlap. E.g. method_[method name]_clustering
.
Project set up
See the dedicated guides for Data modules, Methods modules or Metric modules.
Project population
You can find information about the required metadata (keywords, inputs, outputs, etc) in two ways:
1. Copy the structure of existing modules
Warning
Please do not fork those to avoid transferring unintended metadata from the original project; rather, copy the module structure manually
2. Use the omniValidator
inside your session
The requirements for a given omnibenchmark and keyword can be displayed by running the following lines:
import omniValidator as ov
ov.display_requirements(
benchmark='BENCHMARK_NAME',
keyword='STEP_KEYWORD')
where BENCHMARK_NAME
is the name of the benchmark (omniclustering
) and STEP_KEYWORD
is the keyword associated to the type of module you are working on (omniclustering_dataset
, omniclustering_method
or omniclustering_metric
).
Info
omniValidator
is ported with all projects and doesn't need to be installed.
Output / Omnibenchmark status
An overview of the Omnibenchmark components is available on the Omnibenchmark webpage
The output of the benchmark is available on our shiny app server.