josh.meanings.initializations.utils — org.clojars.joshua/josh.meanings 0.3.4

add-default-chain-length clj

(add-default-chain-length conf)

For monte carlo methods we need a chain length to use when doing sampling. Although callers can pass in a chain length there are some dangers when doing so - for example if the chain length is low it won't necessarily approximate k means plus plus. Meanwhile if the chian length is too low then there will be no point in doing sampling at all - we could just use k means plus plus rather than approximating it.

This function checks to see if a chain length is set and if one is then it does nothing, but it nothing is set it uses the formulas provided in the k means plus plus apporximation papers to ddetermine a reasonable chain length.

centroids->dataset clj

(centroids->dataset conf results)

chain-length-warnings clj

(chain-length-warnings config)

Analyzes the chain length and emits warnings if necessary.

shortest-distance-* clj

(shortest-distance-* configuration)

Denotes the shortest distance from a data point to a center. Which distance to use is decided by the k means configuration.

shortest-distance-squared-* clj

(shortest-distance-squared-* configuration centroids)

Denotes the shortest distance from a data point to a center squared. Useful for computing a D^2 sampling distribution.

should-update-chain-length? clj

(should-update-chain-length? conf)

t-config clj

t-points clj

uniform-sample clj

(uniform-sample ds-seq n & options)

update-chain-length clj

(update-chain-length conf)

weighted-sample clj

(weighted-sample ds-seq weight-fn n & options)