Analyse of file from data.gouv.fr

main.scala

import scala.io.Source
import scala.util.parsing.json.{JSONObject, JSONFormat, JSONArray}

/**
 * Created by fsznajderman on 16/04/2014.
 */
object main {

  case class Accident(lum: Int, agg: Int, int: Int, atm: Int, col: Int, catr: Int, grav: Float)

  case class Result(lbl: String, nb: Int, avg: Float) {
    override def toString: String = s"{ 'title' : $lbl ,'nb' : $nb, 'avg' : $avg} "
  }

  def main(args: Array[String]) {

    val lines = Source.fromFile("/Users/fsznajderman/Documents/Opendata/DataAccidentsCleaned.csv").getLines().toList

    def c(v: String): Int = Option[String](v).filter(s => !s.isEmpty).map(s => s.toInt).getOrElse(0)
    def f(v: String): Float = Option[String](v).filter(s => !s.isEmpty).map(s => s.toFloat).getOrElse(0)

    val allAccidents = lines.map(line => {
      val l = line.split(",")
      Accident(c(l(1)), c(l(2)), c(l(3)), c(l(4)), c(l(5)), c(l(8)), f(l(30)))

    })

    val result = List(('lum, compute(allAccidents.groupBy(i => i.lum))),
      ('agg, compute(allAccidents.groupBy(i => i.agg))),
      ('int, compute(allAccidents.groupBy(i => i.int))),
      ('atm, compute(allAccidents.groupBy(i => i.atm))),
      ('col, compute(allAccidents.groupBy(i => i.col))),
      ('catr, compute(allAccidents.groupBy(i => i.catr))))

    println("{")
    result.foreach(l => formatOutput(l._1, l._2))
    println("}")


  }

  def formatOutput(s: Symbol, l: Iterable[Result]) = {

    println(s"$s':[")
    l.foreach(item => println(item + ","))
    print("],")

  }

  def avg(data: List[Accident]) = data.map(acc => acc.grav).sum / data.size

  def compute(datas: Map[Int, List[Accident]]) = datas.keys.map(k => Result(k.toString, datas(k).size, avg(datas(k))))


}