Initializing playground…
← Back to roadmap

📊 corr & cov — Interactive Playground

pearsonCorr(a, b) computes the Pearson correlation between two Series. dataFrameCorr(df) and dataFrameCov(df) produce symmetric N×N correlation and covariance matrices — mirroring pandas.DataFrame.corr() and pandas.DataFrame.cov().
Edit any code block below and press ▶ Run (or Ctrl+Enter) to execute it live in your browser.

1 · Series pearsonCorr

pearsonCorr(a, b) computes the Pearson correlation coefficient between two Series, aligning on shared index labels and ignoring missing values. Returns a number in [−1, 1], or NaN when a valid result cannot be computed.

TypeScript 
import { Series, pearsonCorr } from "tsb";

const temperature = new Series({ data: [22, 24, 28, 31, 35], name: "temp_C" });
const ice_cream   = new Series({ data: [120, 145, 190, 230, 285], name: "sales" });

const r = pearsonCorr(temperature, ice_cream);
console.log("Pearson r:", r.toFixed(4));   // strong positive correlation

// Negative correlation
const warm_clothes = new Series({ data: [310, 280, 210, 150, 90], name: "jackets" });
console.log("r (temp vs jackets):", pearsonCorr(temperature, warm_clothes).toFixed(4));

// Missing values are dropped per-pair
const c = new Series({ data: [1, null, 3, 4] });
const d = new Series({ data: [2,    4, 6, 8] });
console.log("r (nulls dropped):", pearsonCorr(c, d).toFixed(4));

// Require at least 5 valid pairs — returns NaN when fewer exist
console.log("r (minPeriods=5):", pearsonCorr(c, d, { minPeriods: 5 }));
Click ▶ Run to execute
Ctrl+Enter to run

2 · DataFrame corr matrix (dataFrameCorr)

dataFrameCorr(df) returns a symmetric N×N DataFrame where entry [i, j] is the Pearson correlation between columns i and j. Diagonal entries are always 1. Only numeric columns are included; string columns are silently skipped.

TypeScript 
import { DataFrame, dataFrameCorr } from "tsb";

const df = DataFrame.fromColumns({
  height: [160, 172, 185, 155, 168],
  weight: [55,   72,  90,  48,  65],
  age:    [22,   35,  28,  19,  31],
  city:   ["A", "B", "C", "A", "B"],  // non-numeric — skipped
});

const r = dataFrameCorr(df);
console.log("columns:", [...r.columns.values]);
console.log("shape:", r.shape);

// Read off-diagonal entries via iat() (positional access)
const rHW = r.col("weight").iat(0);  // corr(height, weight)
const rHA = r.col("age").iat(0);     // corr(height, age)
console.log("height–weight r:", rHW.toFixed(4));
console.log("height–age r:   ", rHA.toFixed(4));
console.log("diagonal:", [r.col("height").iat(0), r.col("weight").iat(1), r.col("age").iat(2)]);
Click ▶ Run to execute
Ctrl+Enter to run

3 · DataFrame cov matrix (dataFrameCov)

dataFrameCov(df) returns the sample covariance matrix (denominator n − 1). Pass { ddof: 0 } for population covariance. Diagonal entries are the variance of each column.

TypeScript 
import { DataFrame, dataFrameCov } from "tsb";

const returns = DataFrame.fromColumns({
  AAPL: [ 0.02,  0.01, -0.03,  0.04,  0.01],
  GOOG: [ 0.01,  0.02, -0.02,  0.03,  0.02],
  TSLA: [-0.05,  0.08, -0.10,  0.12, -0.03],
});

const cov = dataFrameCov(returns);
console.log("Covariance matrix:");
for (const col of cov.columns.values) {
  const vals = [...cov.col(col).values].map(v => (v).toFixed(6));
  console.log(col + ":", vals.join("  "));
}

// Compare sample (ddof=1) vs population (ddof=0) variance
const varSample = dataFrameCov(returns, { ddof: 1 }).col("AAPL").iat(0);
const varPop    = dataFrameCov(returns, { ddof: 0 }).col("AAPL").iat(0);
console.log("\nAAPL sample variance:", varSample.toFixed(6));
console.log("AAPL population variance:", varPop.toFixed(6));
Click ▶ Run to execute
Ctrl+Enter to run

🧪 Scratch Pad

Write your own corr & cov code below. All exports from tsb are available: DataFrame, Series, pearsonCorr, dataFrameCorr, dataFrameCov, and more.

TypeScript — Scratch Pad 
import { DataFrame, Series, pearsonCorr, dataFrameCorr, dataFrameCov } from "tsb";

// Try it! Explore correlation and covariance.
const a = new Series({ data: [1, 2, 3, 4, 5] });
const b = new Series({ data: [2, 4, 6, 8, 10] });

console.log("Perfect positive r:", pearsonCorr(a, b));

const df = DataFrame.fromColumns({
  x: [1, 2, 3, 4, 5],
  y: [5, 4, 3, 2, 1],
  z: [2, 4, 6, 8, 10],
});

console.log("\nCorrelation matrix:");
for (const col of dataFrameCorr(df).columns.values) {
  const vals = [...dataFrameCorr(df).col(col).values].map(v => (v).toFixed(2));
  console.log(col + ":", vals.join("  "));
}

console.log("\nCovariance matrix:");
for (const col of dataFrameCov(df).columns.values) {
  const vals = [...dataFrameCov(df).col(col).values].map(v => (v).toFixed(2));
  console.log(col + ":", vals.join("  "));
}
Click ▶ Run to execute
Ctrl+Enter to run