Cascalog data binning code for making histograms

data-binning.clj

(defn one [& _] 1)

(defparallelagg counter
  :init-var one
  :combine-var +)

(defn mn
  [data]
  (<- [?mn]
      (data :> ?v)
      (o/min ?v :> ?mn)))

(defn mx
  [data]
  (<- [?mx]
      (data :> ?v)
      (o/max ?v :> ?mx)))

(defn min-max
  [data]
  (??-
   (mn data)
   (mx data)))

(def num-bins 50)

(defmapfn add-bin
  [mn range d]
  [(-> d
       (- mn)
       (/ range)
       (* num-bins)
       (int)
       (min (dec num-bins)))])

(defn bin
  [data]
  (let [[[[mn]] [[mx]]] (min-max data)
        range (- mx mn)]
    (<- [?bin ?count]
        (data :> ?d)
        (add-bin mn range ?d :> ?bin)
        (counter ?bin :> ?count))))