add precision feature

README.md

@@ -21,6 +21,9 @@ you'll get the correlation coefficient of two columns.
 # check the column is valid
 -c,--check
 
+# set precision, default is 6
+-p,--precision <number>
+
 # show frequency table and histogram
 -b,--binsize <number>
 
@@ -53,7 +56,7 @@ you can download a binary release
 ```sh
 # Install with wget or curl
 ## set the latest version on releases.
-VERSION=v1.0.15
+VERSION=v1.0.20
 ## case you use wget
 wget https://github.com/solaoi/colc/releases/download/$VERSION/colc_linux_amd64.tar.gz
 ## case you use curl
@@ -87,7 +90,6 @@ colc 2 some.csv
 
 <img width="379" alt="スクリーンショット 2022-03-30 14 50 39" src="https://user-images.githubusercontent.com/46414076/160760624-58a15682-0f64-45b0-8b99-2b3732952971.png">
 
-
 Of course `-b,--binsize` works well:)
 
 ```

colc.ts

@@ -7,7 +7,7 @@ import {
 import { runner } from "./lib/common.ts";
 import { parse } from "https://deno.land/[email protected]/flags/mod.ts";
 
-const { _, binsize, b, filter, f, check, c } = parse(Deno.args);
+const { _, binsize, b, filter, f, check, c, precision, p } = parse(Deno.args);
 const [column, filename] = _;
 const hasTwoColumn = (() => {
   if (typeof column !== "string") return false;
@@ -24,6 +24,16 @@ if (
   console.log("Usage:\n  colc [column] [file.csv|tsv|txt]");
   Deno.exit(1);
 }
+const awkPrecision: number = (() => {
+  if (
+    typeof precision === "number" && precision > 0 &&
+    Number.isInteger(precision)
+  ) {
+    return precision;
+  }
+  if (typeof p === "number" && p > 0 && Number.isInteger(p)) return p;
+  return 6;
+})();
 const binSize: number | null = (() => {
   if (typeof binsize === "number" && binsize > 0) return binsize;
   if (typeof b === "number" && b > 0) return b;
@@ -60,7 +70,7 @@ if (hasTwoColumn) {
     }
     bash.push("| awk");
     bash.push(
-      `'BEGIN{OFMT="%.6f"}{split($1,col,",");asum+=col[1];a[NR]=col[1];bsum+=col[2];b[NR]=col[2]}END{amean=asum/NR;bmean=bsum/NR;for(i in a){as+=(a[i]-amean)^2;bs+=(b[i]-bmean)^2;sum+=(a[i]-amean)*(b[i]-bmean)};astddev=sqrt(as/NR);bstddev=sqrt(bs/NR);print sum/NR/astddev/bstddev}'`,
+      `'BEGIN{OFMT="%.${awkPrecision}f"}{split($1,col,",");asum+=col[1];a[NR]=col[1];bsum+=col[2];b[NR]=col[2]}END{amean=asum/NR;bmean=bsum/NR;for(i in a){as+=(a[i]-amean)^2;bs+=(b[i]-bmean)^2;sum+=(a[i]-amean)*(b[i]-bmean)};astddev=sqrt(as/NR);bstddev=sqrt(bs/NR);print sum/NR/astddev/bstddev}'`,
     );
     return bash.join(" ");
   })();
@@ -118,7 +128,7 @@ if (binSize === null) {
     }
     bash.push("| sort -n | awk");
     bash.push(
-      `'BEGIN{OFMT="%.6f"}NR==1{min=$1}{if(0==$1)zeros++;if($1<0)neg++;sum+=$1;d[NR]=$1}END{avg=sum/NR;for(i in d)s+=(d[i]-avg)^2;stddev=sqrt(s/(NR-1));q1=(3*d[int((NR-1)/4)+1]+d[int((NR-1)/4)+2])/4;q3=(d[int(3*(NR-1)/4)+1]+3*d[int(3*(NR-1)/4)+2])/4;iqr=q3-q1;stur=1+log(NR)/log(2);sturi=int(stur);sturges=stur>sturi?sturi+1:sturi;max=d[NR];range=max-min;sqrtnr=sqrt(NR);threerootnr=exp(log(NR)/3);print stddev,avg,sum,NR,max,min,sqrt(s/(NR-1))/sqrtnr,s/(NR-1),(NR%2)?d[(NR+1)/2]:(d[NR/2]+d[NR/2+1])/2,avg+stddev,avg-stddev,avg+2*stddev,avg-2*stddev,avg+3*stddev,avg-3*stddev,range/sturges,(3.5*stddev)/threerootnr,q1,q3,iqr,q1-1.5*iqr,q3+1.5*iqr,range/sqrtnr,2*iqr/threerootnr,range,zeros,zeros*100/NR,neg,neg*100/NR,stddev/avg}'`,
+      `'BEGIN{OFMT="%.${awkPrecision}f"}NR==1{min=$1}{if(0==$1)zeros++;if($1<0)neg++;sum+=$1;d[NR]=$1}END{avg=sum/NR;for(i in d)s+=(d[i]-avg)^2;stddev=sqrt(s/(NR-1));q1=(3*d[int((NR-1)/4)+1]+d[int((NR-1)/4)+2])/4;q3=(d[int(3*(NR-1)/4)+1]+3*d[int(3*(NR-1)/4)+2])/4;iqr=q3-q1;stur=1+log(NR)/log(2);sturi=int(stur);sturges=stur>sturi?sturi+1:sturi;max=d[NR];range=max-min;sqrtnr=sqrt(NR);threerootnr=exp(log(NR)/3);print stddev,avg,sum,NR,max,min,sqrt(s/(NR-1))/sqrtnr,s/(NR-1),(NR%2)?d[(NR+1)/2]:(d[NR/2]+d[NR/2+1])/2,avg+stddev,avg-stddev,avg+2*stddev,avg-2*stddev,avg+3*stddev,avg-3*stddev,range/sturges,(3.5*stddev)/threerootnr,q1,q3,iqr,q1-1.5*iqr,q3+1.5*iqr,range/sqrtnr,2*iqr/threerootnr,range,zeros,zeros*100/NR,neg,neg*100/NR,stddev/avg}'`,
     );
     return bash.join(" ");
   })();
@@ -153,7 +163,7 @@ if (binSize === null) {
     zeroRate,
     negatives,
     negativeRate,
-    cv
+    cv,
   ] = await runner
     .run(statsCommand).then((s) => s.split(" "));
   const sturgesFormulaIsInvalid = count.split(".")[0].length <= 2 &&
@@ -165,8 +175,8 @@ if (binSize === null) {
     "zeros": comma(zeros || "0"),
     "zeros(%)": comma(zeroRate),
     "negatives": comma(negatives || "0"),
-    "negatives(%)": comma(negativeRate)
-  }
+    "negatives(%)": comma(negativeRate),
+  };
   const stats = {
     "min": comma(min),
     "25%(Q1)": comma(q1),
@@ -179,7 +189,7 @@ if (binSize === null) {
     "IQR(Q3-Q1)": comma(iqr),
     "Q1–(1.5*IQR)": comma(lf),
     "Q3+(1.5*IQR)": comma(uf),
-  }
+  };
   const stds = {
     "stddev(σ)": comma(stddev),
     "stderr": comma(stderr),
@@ -188,16 +198,22 @@ if (binSize === null) {
     "mean±σ(≒68%)": `${comma(sigmaMinus1)}, ${comma(sigmaPlus1)}`,
     "mean±2σ(≒95%)": `${comma(sigmaMinus2)}, ${comma(sigmaPlus2)}`,
     "mean±3σ(≒99%)": `${comma(sigmaMinus3)}, ${comma(sigmaPlus3)}`,
-  }
+  };
   const recommendedBinsizes = {
     "binsize(Square-root)": sqrtBinsize,
     ...(!sturgesFormulaIsInvalid && { "binsize(Sturges')": sturgesBinsize }),
     "binsize(Scott's)": scottBinsize,
     "binsize(FD)": fdBinsize,
-  }
+  };
   const { println, showHeader, hr } = formatter(
     sturgesFormulaIsInvalid ? 20 : 21,
-    getMaxLength({...total,...stats,...iqrs,...stds,...recommendedBinsizes}),
+    getMaxLength({
+      ...total,
+      ...stats,
+      ...iqrs,
+      ...stds,
+      ...recommendedBinsizes,
+    }),
   );
   hasHeader ? showHeader(headerName) : hr();
   Object.entries(total).forEach(([key, value]) => {
@@ -236,7 +252,7 @@ const freqCommand = (() => {
   }
   bash.push("| awk");
   bash.push(
-    `'BEGIN{OFMT="%.6f"}{b=int($1/${binSize});a[b]++;bmax=b>bmax?b:bmax;bmin=b<bmin?b:bmin}END{freq="";for(i in a)freq=freq "|" i "_" a[i];print NR, freq, bmin, bmax}'`,
+    `'BEGIN{OFMT="%.${awkPrecision}f"}{b=int($1/${binSize});a[b]++;bmax=b>bmax?b:bmax;bmin=b<bmin?b:bmin}END{freq="";for(i in a)freq=freq "|" i "_" a[i];print NR, freq, bmin, bmax}'`,
   );
   return bash.join(" ");
 })();