Incanter - prepare usage data

usage-chart-preparation.clj

(ns clj-charting.usage-chart-preparation
  (:require
    [incanter.core :refer :all]
    [incanter.stats :as s]
    [incanter.io :as io]))

(defn- resolve-unit-suffix 
  "Replace values such as 333k, 800m, 1.2g with the corresponding value in bytes"
  [val-suffixed]
  (if-let [[_ val unit] (and
                          (string? val-suffixed)
                          (re-find #"(\d+)([kmg])" val-suffixed))]
    (let [order (case unit
                  "k" 1
                  "m" 2
                  "g" 3)
          scale (apply * (take order (repeat 1024)))]
      (* (Integer/parseInt val) scale))))

(defn read-usage-data 
  "Read usage data in the form `sec.ns memory_with_scale_suffix CPU_percentage` into a dataset with
   `ms memory_in_bytes CPU_percentage`"
  [file]
  (let [data (io/read-dataset
               file
               :delim \space)]
    (-> data
        ;; Memory: from 300m or 1g to a number:
        (transform-col
          :col1
          resolve-unit-suffix)
        ;; CPU: From <sec>.<nano> to <ms>:
        (transform-col
          :col0
          #(long (* 1000 %))))))

(defn moving-window-means
  "Given very scattered data, produce a similar sequence of 'moving window mean' where we
   replace each point by the mean of it and the preceding/following `radius` points.
   "
  [radius col]
  (let [x' (concat (repeat radius nil) col)
        parts (drop-last radius (partition (inc (* 2 radius)) 1 x'))
        means (map #(-> (remove nil? %) s/mean long)
                   parts)]
    means))