gistfile1.txt

(defn assert! [p msg] (when-not (p) (throw (Exception. (timbre/info (format msg))))))

(defn jumpoffyr-method-average
  "Takes in a dataset with historical data, a column name to be averaged,
  a name for the average column, and years to start and end averaging for.
  Returns a dataset where the column to be averaged contains the averages
  for all the years of interest and is named to avg-name."
  [historical-data col-to-avg avg-name start-yr-avg end-yr-avg]
  (let [hist-earliest-yr (reduce min (ds/column historical-data :year))
        hist-latest-yr (reduce max (ds/column historical-data :year))
        ;;start-yr (s/validate (s/pred #(>= % hist-earliest-yr)) start-yr-avg)
        end-yr (s/validate (s/pred #(<= % hist-latest-yr)) end-yr-avg)]
    (assert! #(>= start-yr-avg hist-earliest-yr)
             "Start year must be more than or equal to earliest year in dataset")
    (-> (i/query-dataset historical-data
                         {:year {:$gte start-yr-avg
                                 :$lte end-yr}})
        (ds/rename-columns {col-to-avg avg-name})
        (wds/rollup :mean avg-name [:gss-code :sex :age])))
  )