anglican.stat — org.clojars.lsaldyt/anglican 1.0.1

Statistics related functions

collect-by clj

(collect-by f samples)

calculates contribution to log marginal by value;

• accepts a (finite) sequence of samples
• returns a map containing the log sum weight weight for each unique value returned when applying f to a sample, normalized by the total number of samples

collect-predicts clj

(collect-predicts k samples)

constructs a map of weighted predict values

• accepts a key and a (finite) sequence of samples
• returns a map {v log-w} containing an entry v for each unique predict value with key k, along with its log sum weight normalized by the total number of samples

collect-results clj

(collect-results samples)

constructs a map of weighted result values

• accepts a (finite) sequence of samples
• returns a map {v log-w} containing an entry v for each unique predict value along with its log sum weight normalized by the total number of samples

empirical-covariance clj

(empirical-covariance weighted)

calculates the covariance from a collection of weighted pairs [x log-w], where x may be of a number or vector.

empirical-distribution clj

(empirical-distribution weighted)

calculates an empirical distribution from weighted samples;

• accepts a map or sequence of log weighted values [x log-w]
• returns a map {x p} of values and normalized probabilities

empirical-expectation clj

(empirical-expectation f weighted)

calculates an expected value from weighted samples;

• accepts a function f and a map or sequence of log weighted values [x log-w]. f may return either a number or array.
• returns the expected value of (f x)

empirical-kurtosis clj

(empirical-kurtosis weighted)

calculates the standardized kurtosis from a collection of weighted values [x log-w], where x may be of a number or array

empirical-mean clj

(empirical-mean weighted)

calculates the mean from a collection of weighted pairs [x log-w], where x may be of a number or array

empirical-skew clj

(empirical-skew weighted)

calculates the standardized skew from a collection of weighted values [x log-w], where x may be of a number or array

empirical-std clj

(empirical-std weighted)

calculates the standard deviation from a collection of weighted values [x log-w], where x may be of a number or array

empirical-variance clj

(empirical-variance weighted)

calculates the variance from a collection of weighted pairs [x log-w], where x may be of a number or array