taoensso.tufte.stats — com.taoensso/tufte 2.5.0-alpha2

Stats stuff. Private ns, implementation detail.

all-format-columns clj/s

bessel-correction clj/s

(bessel-correction n add)

default-format-columns clj/s

default-format-id-fn clj/s

doubles? clj

(doubles? x)

Returns true iff given double array

get-max-id-width clj/s

(get-max-id-width id-sstats*
                  {:keys [format-id-fn]
                   :or {format-id-fn default-format-id-fn}})

is-p clj/s

(is-p x)

ISummaryStatsBuffered clj/s protocol

ssb-clear clj/s

(ssb-clear _)

Clears all internal state and returns nil.

ssb-deref clj/s

(ssb-deref _)

(ssb-deref _ flush-buffer?)

Returns current ?sstats.

ssb-flush clj/s

(ssb-flush _)

Flushes internal buffer and returns newly merged sstats or nil.

ssb-push clj/s

(ssb-push _ n)

Adds given num to internal buffer.

longs? clj

(longs? x)

Returns true iff given long array

multi-reduce clj/s

(multi-reduce f init coll)

(multi-reduce f1 init1 f2 init2 coll)

(multi-reduce f1 init1 f2 init2 f3 init3 coll)

Like reduce but supports separate simultaneous accumulators as a micro-optimisation when reducing a large collection multiple times.

percentile clj/s

(percentile nums p)

Returns ?double

percentiles clj/s

(percentiles nums)

Returns ?[min p25 p50 p75 p90 p95 p99 max] doubles in:

• O(1) for Sorted types (SortedLongs, SortedDoubles),
• O(n.log_n) otherwise.

rf-sum clj/s

(rf-sum acc in)

rf-sum-abs-deviation clj/s

(rf-sum-abs-deviation central-point acc x)

rf-sum-variance clj/s

(rf-sum-variance xbar acc x)

sorted-doubles clj/s

(sorted-doubles nums)

sorted-doubles? clj/s

(sorted-doubles? x)

summary-stats clj/s

(summary-stats nums-or-ss-map)

Given a coll of numbers or previously dereffed SummaryStats map, returns a new mergeable ?SummaryStats with: (deref ss) => {:keys [n min max p25 ... p99 mean var mad]}

See also summary-stats-merge.

summary-stats-buffered clj/s

(summary-stats-buffered)

(summary-stats-buffered {:keys [buffer-size buffer-init sstats-init merge-cb]
                         :or {buffer-size 100000.0}
                         :as opts})

Returns a new stateful SummaryStatsBuffered with: (ssb <num>) => Adds given number to internal buffer. (deref ssb) => Flushes buffer if necessary, and returns a mergeable ?SummaryStats. Deref again to get a map of summary stats for all numbers ever added to ssb: {:keys [n min max p25 ... p99 mean var mad]}.

Useful for summarizing a (possibly infinite) stream of numbers.

Options: :buffer-size - The maximum number of numbers that may be buffered before next (ssb <num>) call will block to flush buffer and merge with any existing summary stats.

             Larger buffers mean better performance and more
             accurate stats, at the cost of more memory use
             while buffering.

:buffer-init - Initial buffer content, useful for persistent ssb. :sstats-init - Initial summary stats, useful for persistent ssb.

summary-stats-buffered-fast clj/s

(summary-stats-buffered-fast buffer-size merge-cb)

Returns fastest possible SummaryStatsBuffered.

summary-stats-buffered? clj/s

(summary-stats-buffered? x)

Returns true iff given a SummaryStatsBuffered instance.

summary-stats-clear! clj/s

(summary-stats-clear! stateful-summary-stats)

Clears internal state (incl. stats and buffers, etc.) for given stateful SummaryStats instance and returns nil.

summary-stats-format clj/s

(summary-stats-format clock-total
                      id-sstats*
                      {:keys [columns sort-fn format-id-fn max-id-width]
                       :as opts
                       :or {columns default-format-columns
                            sort-fn (fn [ss] (get (enc/force-ref ss) :sum))
                            format-id-fn default-format-id-fn}})

Given {<id> <sstats>} or {<id> <sstats-map>}, returns a formatted table string. Assumes nanosecond clock, and stats based on profiling id'd nanosecond times.

summary-stats-merge clj/s

(summary-stats-merge ss1)

(summary-stats-merge ss1 ss2)

Given one or more SummaryStats, returns a new ?SummaryStats with: (summary-stats-merge (summary-stats nums1) (summary-stats nums2))

an approximatation of (summary-stats (merge nums1 nums2))

Useful when you want summary stats for a large coll of numbers for which it would be infeasible/expensive to keep all numbers for accurate merging.

summary-stats-stateful? clj/s

(summary-stats-stateful? x)

Returns true iff given a stateful SummaryStats instance.

summary-stats? clj/s

(summary-stats? x)

Returns true iff given a SummaryStats argument.