clj-ml.attribute-selection — cc.artifice/clj-ml 0.8.7

apply-opts clj

(apply-opts f opts)

attribute-eval-options clj

(attribute-eval-options m)

attribute-selector clj

(attribute-selector &
                    {:keys [search evaluator folds random-seed rank] :as opts})

Returns an attibute selector. A search and evaluator object are required. You may also specify that a cross-validation needs to be done by passing in the number of folds.

best-first clj

(best-first & {:as options})

cfs-subset-eval clj

(cfs-subset-eval & {:as options})

chi-squared clj

(chi-squared & {:as options})

desc clj

(desc a b)

gain-ratio clj

(gain-ratio & {:as options})

genetic clj

(genetic & {:as options})

greedy clj

(greedy & {:as options})

info-gain clj

(info-gain & {:as options})

make-weka-obj clj

(make-weka-obj kind options)

obj-aliases clj

Mapping of cjl-ml keywords to actual Weka classes

one-R clj

(one-R & {:as options})

rank-attributes clj

(rank-attributes ds evaluator)

Similar to what select-attributes with ranker would do but it returns the score associated with each attribute.

rank-search clj

(rank-search & {:keys [evaluator step-size start-point] :as opts})

ranker clj

(ranker & {:as options})

relief clj

(relief & {:as options})

select-attributes clj

(select-attributes ds & {:keys [selector search evaluator] :as opts})

Takes a data set and a selector, or options to build a selector. The best attributes are then selected using the selector's settings. A list of attributes will be returned along with the used selector in the metadata of the list.

symmetrical-uncert clj

(symmetrical-uncert & {:as options})