clj-nd4j.dataset — io.github.jeremylcfr/clj-nd4j 0.1.0-SNAPSHOT

->dataset clj

(->dataset)

(->dataset obj)

(->dataset features labels)

(->dataset features labels features-mask labels-mask)

Generic dataset coercer. Uses ->nd-array to coerce input which can be so a clojure structure. Single-arity version does nothing if input is a DataSet and parses a clojure structure if valid (note that it is the less performant way of doing this) You can pass a map : {:features ... :labels ... :features-mask ... :labels-mask ...} or a nested/empty vector (in same order) [[ ... ] [ ... ] [ ... ] [ ... ]] or nil (~ no arg) Valid args are either :

• no arg
• two args (features and labels)
• four args (all but masks are nilable)

->immutable-transformation1 clj

(->immutable-transformation1 transform-fn)

->immutable-transformation2 clj

(->immutable-transformation2 transform-fn)

binarize clj

(binarize data)

(binarize data cutoff)

binarize! clj

(binarize! data)

(binarize! data cutoff)

clj->dataset clj multimethod

copy clj

(copy data)

dataset clj

(dataset)

(dataset features labels)

(dataset features labels features-mask labels-mask)

Strict wrapper of DataSet constructors, i.e. accepts only NDArray instances as input. Use ->dataset for a more flexible (but a bit less performant) coercer

dataset? clj

(dataset? obj)

deep-split-test-and-train clj

(deep-split-test-and-train dataset num-samples)

(deep-split-test-and-train dataset num-samples random)

do-binarize! clj

(do-binarize! data)

(do-binarize! data cutoff)

do-multiply! clj

(do-multiply! data multiplicator)

do-normalize! clj

(do-normalize! data)

do-scale! clj

(do-scale! data)

do-shuffle! clj

(do-shuffle! data)

extract-features-and-labels clj

(extract-features-and-labels dataset)

load-dataset clj

(load-dataset io-coercible)

Loads a dataset from a file written preferably with write-dataset!

multiply clj

multiply! clj

(multiply! data multiplicator)

normalize clj

normalize! clj

(normalize! data)

raw-split-test-and-train clj

(raw-split-test-and-train dataset num-samples)

(raw-split-test-and-train dataset num-samples seed-or-random)

read-dataset clj

(read-dataset io-coercible)

(read-dataset io-coercible data)

read-dataset! clj

(read-dataset! io-coercible)

(read-dataset! io-coercible data)

scale clj

scale! clj

(scale! data)

shuffle clj

shuffle! clj

(shuffle! data)

split-test-and-train clj

(split-test-and-train dataset num-samples)

(split-test-and-train dataset num-samples seed-or-random)

write-dataset! clj

(write-dataset! io-coercible data)

Writes a dataset to a file which can be read back with load-dataset! saving some clojure interop loss