F# - 基于百分位比较用标签替换浮点数

Question

我需要根据与 prcntls 中的百分位数的比较，将 tmp 中的值映射到 tmpRplcd 中的标签，但是 Array .map2 行失败，因为数组的长度不同。

module SOQN = 
   open System
   open MathNet.Numerics.Statistics
   let tmp = [| 13.0; 17.0; 23.0; 11.0; 11.0; 13.0; 31.0; 
                19.0; 47.0; 29.0; 29.0; 19.0; 43.0; 37.0 |]
   let tmpPrcntls = 
      tmp
      |> Array.sort
   let lbls = [| "p0"; "p1"; "p2"; "p3"; "p4";"p5" |]
   let prcntls = [| Statistics.Percentile(tmpPrcntls,0)   // 11.0
                    Statistics.Percentile(tmpPrcntls,20)  // 13.0
                    Statistics.Percentile(tmpPrcntls,40)  // 19.0
                    Statistics.Percentile(tmpPrcntls,60)  // 28.6
                    Statistics.Percentile(tmpPrcntls,80)  // 35.8
                    Statistics.Percentile(tmpPrcntls,100) // 47.0
                 |]
   let lkpTbl = Map(Array.zip prcntls lbls)
   let tmpRplcd:string[] = 
      tmp
      |> Array.map2 (fun x y -> if x <= y then lkpTbl.[y] else "") prcntls
   let main = 
      printfn ""
      printfn "Percentile Test"
      printfn ""
      printfn "tmpPrcntls: %A" tmpPrcntls
      printfn "prcntls:%A" prcntls
      printfn "tmpRplcd:%A" tmpRplcd
      0
   [<EntryPoint>]
   main
   |> ignore

 // Expected Result:
 // tmpRplcd = [| "p1"; "p2"; "p3"; "p0"; "p0"; "p1"; "p4"; 
 //               "p2"; "p5"; "p4"; "p4"; "p2"; "p5"; "p5" |]   

我哪里错了？

Answer 1

以下是您打算对 F# 程序执行的操作。

我从http://www.dummies.com/education/math/statistics/how-to-calculate-percentiles-in-statistics/ 提出了一个百分位数计算的实现，它显示在下面的统计模块中。

namespace FSharpBasics

module Statistics =
    let percentile p (array: float[]) =
        let threshold = (float p / 100.0) * float (array |> Array.length)
        let thresholdCeiling = int (System.Math.Ceiling threshold)
        let thresholdInteger = int (threshold)
        array
            |> Array.sort
            |> Array.skip (thresholdCeiling - 1)
            |> Array.truncate (if thresholdInteger = thresholdCeiling then 2 else 1)
            |> Array.average

module PercentileTest =
    open System

    let tmp = [| 13.0; 17.0; 23.0; 11.0; 11.0; 13.0; 31.0;
                 19.0; 47.0; 29.0; 29.0; 19.0; 43.0; 37.0 |]

    let lbls = 
        [| for n in 0..20..100 -> "p" + string (n / 20) |]

    let prcntls = 
        [| for n in 0..20..100 -> Statistics.percentile n tmp |]

    let tmpPrcntls = 
        tmp |> Array.sort

    let lkpTbl = 
        Array.zip prcntls lbls

    let tmpRplcd : string[] =
        tmp
        |> Array.map (fun x -> 
                lkpTbl 
                |> Array.filter (fun (prcntl, lbl) -> prcntl <= x)
                |> Array.last
                |> snd)

    [<EntryPoint>]
    let main argv =
        printfn ""
        printfn "Percentile Test"
        printfn ""
        printfn "tmp: %A" tmp
        printfn "tmpPrcntls: %A" tmpPrcntls
        printfn "prcntls: %A" prcntls
        printfn "tmpRplcd: %A" tmpRplcd
        System.Console.ReadKey() |> ignore
        0 // return an integer exit code
(*---- output ----

Percentile Test

tmp: [|13.0; 17.0; 23.0; 11.0; 11.0; 13.0; 31.0; 19.0; 47.0; 29.0; 29.0; 19.0; 43.0;
  37.0|]
tmpPrcntls: [|11.0; 11.0; 13.0; 13.0; 17.0; 19.0; 19.0; 23.0; 29.0; 29.0; 31.0; 37.0; 43.0;
  47.0|]
prcntls: [|11.0; 13.0; 19.0; 29.0; 37.0; 47.0|]
tmpRplcd: [|"p1"; "p1"; "p2"; "p0"; "p0"; "p1"; "p3"; "p2"; "p5"; "p3"; "p3"; "p2"; "p4";
 "p4"|]
---- ----*)

Answer 2

我认为您对 map2 的使用是错误的 - map2 函数压缩两个数组，然后将给定函数应用于压缩数组。

根据您的问题，我的猜测是您实际上想做其他事情。对于每个输入，您都希望遍历所有百分位数并找到第一个百分位数，使得该值大于（或小于？）该百分位数。为此，您需要将map2 替换为以下内容：

let tmpRplcd:string[] = 
  tmp 
  |> Array.map (fun y -> 
    prcntls |> Array.tryPick (fun x ->
      if x <= y then Some(lkpTbl.[x]) else None))
  |> Array.map (fun v -> defaultArg v "")

我没有合适的版本来尝试这个，但我认为这应该可以满足您的需要（我只是不确定您是否需要 x <= y 或相反！）