gistfile1.clj

(defn prob-n [pred tol in n] ;; in is a csv's filename
  (let [as (for [[_ _ p m] (map #(.split % ";") (rest (.split (slurp in) "\n"))) :when p]
             (if (empty? m) [p] [p m])) ;; take the third and fourth (if possible) column of each row
        conteo (for [as (c/combinations as n)]
                 (if (apply pred tol as) 1 0))]
    (/ (reduce + conteo) (count conteo))))

(defn ningun-apellido-en-comun [tol & as]
  (let [distinct-as (distinct (apply concat as))
        conteo (for [a distinct-as]
                 (count (filter #(some #{a} %) as)))]
    (<= (- (reduce + conteo) (count distinct-as)) tol)))

user=> (prob-n ningun-apellido-en-comun 0 "some.csv" 20)

gistfile2.clj

(defn prob-n [pred tol in n]
  (let [as (for [line (rest (.split (slurp in) "\n"))
                 :let [as (distinct (filter seq (take 2 (drop 2 (.split line ";")))))]
                 :when (seq as)] as)
        conteo (for [as (c/combinations as n)]
                 (if (apply pred tol as) 1 0))]
    (/ (reduce + conteo) (count conteo))))

(defn ningun-apellido-en-comun [tol & as]
  (let [freqs (frequencies as)]
    (<= (- (reduce + (vals freqs)) (count (keys freqs))) tol)))

user=> (prob-n ningun-apellido-en-comun 0 "some.csv" 20)