Module 9 · Dependence, Regression, and Model Foundations Lesson 85 of 120
Correlation, Causation, Confounding, and Spurious Relationships
A pooled recovery-rate comparison that reverses within difficulty groups.
Transcript
19 sentences · select one to jump thereCheck your understanding
Does this grouped example prove a causal benefit from switching every case to B?
Code lab
Run it yourself
The lesson source in 7 languages. Edit it, run TypeScript and Python right here, and compare with the expected output.
/**
* Fintech Math Bootcamp · Lesson 085 of 120
* Correlation, Causation, Confounding, and Spurious Relationships
* Module 09: Dependence, Regression, and Model Foundations
*
* Scenario: A pooled recovery-rate comparison that reverses within difficulty groups
* Rule: association ≠ effect of intervention
*
* Try it: Does this grouped example prove a causal benefit from switching every case to B?
*
* Lesson article: https://thefintechbuilder.com/financial-mathematics-statistics-and-data-foundations/dependence-regression-and-model-foundations/correlation-causation-confounding-and-spurious-relationships/
* Free course: https://courses.thefintechbuilder.com
* Synthetic teaching example, not financial advice or a production library.
*/
export function lesson085() {
const a={easyRecovered:90,easy:100,hardRecovered:1,hard:10};
const b={easyRecovered:19,easy:20,hardRecovered:20,hard:100};
const result={pooledA:(a.easyRecovered+a.hardRecovered)/(a.easy+a.hard),
pooledB:(b.easyRecovered+b.hardRecovered)/(b.easy+b.hard),
easy:[a.easyRecovered/a.easy,b.easyRecovered/b.easy],
hard:[a.hardRecovered/a.hard,b.hardRecovered/b.hard]};
return result;
}
export const checkedResult = {"pooledA":0.8272727272727273,"pooledB":0.325,"easy":[0.9,0.95],"hard":[0.1,0.2]};
// Run this file directly: npx tsx lessons/09-dependence-regression-and-model-foundations/085-correlation-causation-confounding-and-spurious-relationships.ts
if (process.argv[1] && import.meta.url.endsWith(process.argv[1].replace(/\\/g, "/").split("/").pop()!)) {
console.log(JSON.stringify(lesson085(), null, 2));
}
Your output
Press Run to execute the code in your browser.
Expected output
{
"pooledA": 0.8272727272727273,
"pooledB": 0.325,
"easy": [
0.9,
0.95
],
"hard": [
0.1,
0.2
]
}"""
Fintech Math Bootcamp · Lesson 085 of 120
Correlation, Causation, Confounding, and Spurious Relationships
Module 09: Dependence, Regression, and Model Foundations
Scenario: A pooled recovery-rate comparison that reverses within difficulty groups
Rule: association ≠ effect of intervention
Try it: Does this grouped example prove a causal benefit from switching every case to B?
Lesson article: https://thefintechbuilder.com/financial-mathematics-statistics-and-data-foundations/dependence-regression-and-model-foundations/correlation-causation-confounding-and-spurious-relationships/
Free course: https://courses.thefintechbuilder.com
Synthetic teaching example, not financial advice or a production library.
"""
import json
def lesson_085():
a = {"easy_recovered": 90, "easy": 100, "hard_recovered": 1, "hard": 10}
b = {"easy_recovered": 19, "easy": 20, "hard_recovered": 20, "hard": 100}
return {
"pooledA": (a["easy_recovered"] + a["hard_recovered"]) / (a["easy"] + a["hard"]),
"pooledB": (b["easy_recovered"] + b["hard_recovered"]) / (b["easy"] + b["hard"]),
"easy": [a["easy_recovered"] / a["easy"], b["easy_recovered"] / b["easy"]],
"hard": [a["hard_recovered"] / a["hard"], b["hard_recovered"] / b["hard"]],
}
if __name__ == "__main__":
print(json.dumps(lesson_085(), indent=2))
Your output
Press Run to execute the code in your browser.
Expected output
{
"pooledA": 0.8272727272727273,
"pooledB": 0.325,
"easy": [
0.9,
0.95
],
"hard": [
0.1,
0.2
]
}// Fintech Math Bootcamp - Lesson 085 of 120
// Correlation, Causation, Confounding, and Spurious Relationships
// Module 09: Dependence, Regression, and Model Foundations
//
// Scenario: A pooled recovery-rate comparison that reverses within difficulty groups
// Rule: association != effect of intervention
//
// Try it: Does this grouped example prove a causal benefit from switching every case to B?
//
// Lesson article: https://thefintechbuilder.com/financial-mathematics-statistics-and-data-foundations/dependence-regression-and-model-foundations/correlation-causation-confounding-and-spurious-relationships/
// Free course: https://courses.thefintechbuilder.com
// Synthetic teaching example, not financial advice or a production library.
import java.util.ArrayList;
import java.util.Arrays;
import java.util.LinkedHashMap;
import java.util.List;
import java.util.Map;
public class Main {
// Recovered and total case counts by difficulty.
static final class Outcomes {
final double easyRecovered, easy, hardRecovered, hard;
Outcomes(double easyRecovered, double easy, double hardRecovered, double hard) {
this.easyRecovered = easyRecovered;
this.easy = easy;
this.hardRecovered = hardRecovered;
this.hard = hard;
}
}
static Map<String, Object> lesson085() {
Outcomes a = new Outcomes(90, 100, 1, 10);
Outcomes b = new Outcomes(19, 20, 20, 100);
Map<String, Object> result = new LinkedHashMap<String, Object>();
result.put("pooledA", (a.easyRecovered + a.hardRecovered) / (a.easy + a.hard));
result.put("pooledB", (b.easyRecovered + b.hardRecovered) / (b.easy + b.hard));
result.put("easy", new double[] {a.easyRecovered / a.easy, b.easyRecovered / b.easy});
result.put("hard", new double[] {a.hardRecovered / a.hard, b.hardRecovered / b.hard});
return result;
}
public static void main(String[] args) {
System.out.println(toJson(lesson085(), ""));
}
// Minimal JSON writer: two-space indent, whole numbers without a decimal point, NaN as null.
static String toJson(Object value, String indent) {
if (value == null) return "null";
if (value instanceof Boolean) return value.toString();
if (value instanceof Number) return formatNumber(((Number) value).doubleValue());
if (value instanceof String) return quote((String) value);
if (value instanceof double[]) {
List<Object> boxed = new ArrayList<Object>();
for (double d : (double[]) value) boxed.add(d);
return toJson(boxed, indent);
}
if (value instanceof Object[]) return toJson(Arrays.asList((Object[]) value), indent);
String inner = indent + " ";
StringBuilder out = new StringBuilder();
if (value instanceof Map) {
Map<?, ?> map = (Map<?, ?>) value;
if (map.isEmpty()) return "{}";
out.append("{\n");
int i = 0;
for (Map.Entry<?, ?> entry : map.entrySet()) {
out.append(inner).append(quote(entry.getKey().toString())).append(": ")
.append(toJson(entry.getValue(), inner));
out.append(++i < map.size() ? ",\n" : "\n");
}
return out.append(indent).append("}").toString();
}
List<?> list = (List<?>) value;
if (list.isEmpty()) return "[]";
out.append("[\n");
for (int i = 0; i < list.size(); i++) {
out.append(inner).append(toJson(list.get(i), inner));
out.append(i + 1 < list.size() ? ",\n" : "\n");
}
return out.append(indent).append("]").toString();
}
static String formatNumber(double x) {
if (Double.isNaN(x) || Double.isInfinite(x)) return "null";
if (x == Math.rint(x) && Math.abs(x) < 1e15) return Long.toString((long) x);
return Double.toString(x);
}
static String quote(String s) {
StringBuilder out = new StringBuilder("\"");
for (char c : s.toCharArray()) {
if (c == '"' || c == '\\') out.append('\\').append(c);
else if (c == '\n') out.append("\\n");
else if (c < 0x20) out.append(String.format("\\u%04x", (int) c));
else out.append(c);
}
return out.append('"').toString();
}
}
No browser runner for Java yet
Read the code here, then run it in your own toolchain or a ready-made cloud workspace.
Expected output
{
"pooledA": 0.8272727272727273,
"pooledB": 0.325,
"easy": [
0.9,
0.95
],
"hard": [
0.1,
0.2
]
}// Fintech Math Bootcamp · Lesson 085 of 120
// Correlation, Causation, Confounding, and Spurious Relationships
// Module 09: Dependence, Regression, and Model Foundations
//
// Scenario: A pooled recovery-rate comparison that reverses within difficulty groups
// Rule: association ≠ effect of intervention
//
// Try it: Does this grouped example prove a causal benefit from switching every case to B?
//
// Lesson article: https://thefintechbuilder.com/financial-mathematics-statistics-and-data-foundations/dependence-regression-and-model-foundations/correlation-causation-confounding-and-spurious-relationships/
// Free course: https://courses.thefintechbuilder.com
// Synthetic teaching example, not financial advice or a production library.
package main
import (
"encoding/json"
"fmt"
)
type Lesson085Result struct {
PooledA float64 `json:"pooledA"`
PooledB float64 `json:"pooledB"`
Easy []float64 `json:"easy"`
Hard []float64 `json:"hard"`
}
// Outcomes holds recovered and total case counts by difficulty.
type Outcomes struct {
EasyRecovered, Easy, HardRecovered, Hard float64
}
func lesson085() Lesson085Result {
a := Outcomes{EasyRecovered: 90, Easy: 100, HardRecovered: 1, Hard: 10}
b := Outcomes{EasyRecovered: 19, Easy: 20, HardRecovered: 20, Hard: 100}
return Lesson085Result{
PooledA: (a.EasyRecovered + a.HardRecovered) / (a.Easy + a.Hard),
PooledB: (b.EasyRecovered + b.HardRecovered) / (b.Easy + b.Hard),
Easy: []float64{a.EasyRecovered / a.Easy, b.EasyRecovered / b.Easy},
Hard: []float64{a.HardRecovered / a.Hard, b.HardRecovered / b.Hard},
}
}
func main() {
out, err := json.MarshalIndent(lesson085(), "", " ")
if err != nil {
panic(err)
}
fmt.Println(string(out))
}
No browser runner for Go yet
Read the code here, then run it in your own toolchain or a ready-made cloud workspace.
Expected output
{
"pooledA": 0.8272727272727273,
"pooledB": 0.325,
"easy": [
0.9,
0.95
],
"hard": [
0.1,
0.2
]
}// Fintech Math Bootcamp · Lesson 085 of 120
// Correlation, Causation, Confounding, and Spurious Relationships
// Module 09: Dependence, Regression, and Model Foundations
//
// Scenario: A pooled recovery-rate comparison that reverses within difficulty groups
// Rule: association ≠ effect of intervention
//
// Try it: Does this grouped example prove a causal benefit from switching every case to B?
//
// Lesson article: https://thefintechbuilder.com/financial-mathematics-statistics-and-data-foundations/dependence-regression-and-model-foundations/correlation-causation-confounding-and-spurious-relationships/
// Free course: https://courses.thefintechbuilder.com
// Synthetic teaching example, not financial advice or a production library.
#include <cmath>
#include <cstdio>
#include <cstdlib>
#include <iostream>
#include <optional>
#include <stdexcept>
#include <string>
#include <utility>
#include <vector>
// A minimal JSON value, enough to print this lesson's result.
struct Json {
enum class Kind { Null, Bool, Number, String, Array, Object };
Kind kind = Kind::Null;
bool flag = false;
double number = 0.0;
std::string text;
std::vector<std::string> keys; // object keys, parallel to items
std::vector<Json> items; // array elements or object values
Json() = default;
Json(bool value) : kind(Kind::Bool), flag(value) {}
Json(int value) : kind(Kind::Number), number(value) {}
Json(double value) : kind(Kind::Number), number(value) {}
Json(const char* value) : kind(Kind::String), text(value) {}
Json(const std::string& value) : kind(Kind::String), text(value) {}
Json(const std::vector<double>& values) : kind(Kind::Array) {
for (double v : values) items.push_back(Json(v));
}
};
Json jsonArray(const std::vector<Json>& values) {
Json array;
array.kind = Json::Kind::Array;
array.items = values;
return array;
}
Json jsonObject(const std::vector<std::pair<std::string, Json>>& fields) {
Json object;
object.kind = Json::Kind::Object;
for (const auto& field : fields) {
object.keys.push_back(field.first);
object.items.push_back(field.second);
}
return object;
}
// Shortest decimal form that reads back as the same double.
std::string formatNumber(double x) {
if (!std::isfinite(x)) return "null";
char buffer[32];
if (x == std::floor(x) && std::fabs(x) < 1e15) {
std::snprintf(buffer, sizeof buffer, "%.0f", x);
return buffer;
}
for (int precision = 1; precision <= 17; ++precision) {
std::snprintf(buffer, sizeof buffer, "%.*g", precision, x);
if (std::strtod(buffer, nullptr) == x) break;
}
return buffer;
}
std::string quote(const std::string& s) {
std::string out = "\"";
for (char c : s) {
if (c == '"' || c == '\\') { out += '\\'; out += c; }
else if (c == '\n') out += "\\n";
else out += c;
}
return out + "\"";
}
std::string toJson(const Json& value, const std::string& indent = "") {
switch (value.kind) {
case Json::Kind::Null: return "null";
case Json::Kind::Bool: return value.flag ? "true" : "false";
case Json::Kind::Number: return formatNumber(value.number);
case Json::Kind::String: return quote(value.text);
default: break;
}
const bool isObject = value.kind == Json::Kind::Object;
if (value.items.empty()) return isObject ? "{}" : "[]";
const std::string inner = indent + " ";
std::string out = isObject ? "{\n" : "[\n";
for (std::size_t i = 0; i < value.items.size(); ++i) {
out += inner;
if (isObject) out += quote(value.keys[i]) + ": ";
out += toJson(value.items[i], inner);
out += i + 1 < value.items.size() ? ",\n" : "\n";
}
return out + indent + (isObject ? "}" : "]");
}
// Recovered and total case counts by difficulty.
struct Outcomes {
double easyRecovered, easy, hardRecovered, hard;
};
Json lesson085() {
const Outcomes a{90, 100, 1, 10};
const Outcomes b{19, 20, 20, 100};
return jsonObject({
{"pooledA", (a.easyRecovered + a.hardRecovered) / (a.easy + a.hard)},
{"pooledB", (b.easyRecovered + b.hardRecovered) / (b.easy + b.hard)},
{"easy", std::vector<double>{a.easyRecovered / a.easy, b.easyRecovered / b.easy}},
{"hard", std::vector<double>{a.hardRecovered / a.hard, b.hardRecovered / b.hard}},
});
}
int main() {
std::cout << toJson(lesson085()) << '\n';
return 0;
}
No browser runner for C++ yet
Read the code here, then run it in your own toolchain or a ready-made cloud workspace.
Expected output
{
"pooledA": 0.8272727272727273,
"pooledB": 0.325,
"easy": [
0.9,
0.95
],
"hard": [
0.1,
0.2
]
}// Fintech Math Bootcamp · Lesson 085 of 120
// Correlation, Causation, Confounding, and Spurious Relationships
// Module 09: Dependence, Regression, and Model Foundations
//
// Scenario: A pooled recovery-rate comparison that reverses within difficulty groups
// Rule: association ≠ effect of intervention
//
// Try it: Does this grouped example prove a causal benefit from switching every case to B?
//
// Lesson article: https://thefintechbuilder.com/financial-mathematics-statistics-and-data-foundations/dependence-regression-and-model-foundations/correlation-causation-confounding-and-spurious-relationships/
// Free course: https://courses.thefintechbuilder.com
// Synthetic teaching example, not financial advice or a production library.
/// A minimal JSON value, enough to print this lesson's result.
#[allow(dead_code)]
enum Json {
Null,
Bool(bool),
Num(f64),
Str(String),
Arr(Vec<Json>),
Obj(Vec<(String, Json)>),
}
#[allow(dead_code)]
impl Json {
fn obj(fields: Vec<(&str, Json)>) -> Json {
Json::Obj(fields.into_iter().map(|(k, v)| (k.to_string(), v)).collect())
}
fn nums(values: &[f64]) -> Json {
Json::Arr(values.iter().map(|&v| Json::Num(v)).collect())
}
/// Pretty-prints with two-space indentation.
fn pretty(&self, indent: &str) -> String {
let inner = format!("{} ", indent);
match self {
Json::Null => "null".to_string(),
Json::Bool(b) => b.to_string(),
Json::Num(x) => format_number(*x),
Json::Str(s) => quote(s),
Json::Arr(items) if items.is_empty() => "[]".to_string(),
Json::Obj(fields) if fields.is_empty() => "{}".to_string(),
Json::Arr(items) => {
let body: Vec<String> = items
.iter()
.map(|v| format!("{}{}", inner, v.pretty(&inner)))
.collect();
format!("[\n{}\n{}]", body.join(",\n"), indent)
}
Json::Obj(fields) => {
let body: Vec<String> = fields
.iter()
.map(|(k, v)| format!("{}{}: {}", inner, quote(k), v.pretty(&inner)))
.collect();
format!("{{\n{}\n{}}}", body.join(",\n"), indent)
}
}
}
}
fn format_number(x: f64) -> String {
if !x.is_finite() {
"null".to_string()
} else if x == x.trunc() && x.abs() < 1e15 {
format!("{}", x as i64)
} else {
format!("{}", x)
}
}
fn quote(s: &str) -> String {
let mut out = String::from("\"");
for c in s.chars() {
match c {
'"' => out.push_str("\\\""),
'\\' => out.push_str("\\\\"),
'\n' => out.push_str("\\n"),
c => out.push(c),
}
}
out.push('"');
out
}
/// Recovered and total case counts by difficulty.
struct Outcomes {
easy_recovered: f64,
easy: f64,
hard_recovered: f64,
hard: f64,
}
fn lesson_085() -> Json {
let a = Outcomes { easy_recovered: 90.0, easy: 100.0, hard_recovered: 1.0, hard: 10.0 };
let b = Outcomes { easy_recovered: 19.0, easy: 20.0, hard_recovered: 20.0, hard: 100.0 };
Json::obj(vec![
("pooledA", Json::Num((a.easy_recovered + a.hard_recovered) / (a.easy + a.hard))),
("pooledB", Json::Num((b.easy_recovered + b.hard_recovered) / (b.easy + b.hard))),
("easy", Json::nums(&[a.easy_recovered / a.easy, b.easy_recovered / b.easy])),
("hard", Json::nums(&[a.hard_recovered / a.hard, b.hard_recovered / b.hard])),
])
}
fn main() {
println!("{}", lesson_085().pretty(""));
}
No browser runner for Rust yet
Read the code here, then run it in your own toolchain or a ready-made cloud workspace.
Expected output
{
"pooledA": 0.8272727272727273,
"pooledB": 0.325,
"easy": [
0.9,
0.95
],
"hard": [
0.1,
0.2
]
}// Fintech Math Bootcamp · Lesson 085 of 120
// Correlation, Causation, Confounding, and Spurious Relationships
// Module 09: Dependence, Regression, and Model Foundations
//
// Scenario: A pooled recovery-rate comparison that reverses within difficulty groups
// Rule: association ≠ effect of intervention
//
// Try it: Does this grouped example prove a causal benefit from switching every case to B?
//
// Lesson article: https://thefintechbuilder.com/financial-mathematics-statistics-and-data-foundations/dependence-regression-and-model-foundations/correlation-causation-confounding-and-spurious-relationships/
// Free course: https://courses.thefintechbuilder.com
// Synthetic teaching example, not financial advice or a production library.
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text.Json;
var options = new JsonSerializerOptions { WriteIndented = true };
Console.WriteLine(JsonSerializer.Serialize(Lesson085(), options));
static object Lesson085()
{
var a = new { easyRecovered = 90.0, easy = 100.0, hardRecovered = 1.0, hard = 10.0 };
var b = new { easyRecovered = 19.0, easy = 20.0, hardRecovered = 20.0, hard = 100.0 };
return new
{
pooledA = (a.easyRecovered + a.hardRecovered) / (a.easy + a.hard),
pooledB = (b.easyRecovered + b.hardRecovered) / (b.easy + b.hard),
easy = new[] { a.easyRecovered / a.easy, b.easyRecovered / b.easy },
hard = new[] { a.hardRecovered / a.hard, b.hardRecovered / b.hard },
};
}
No browser runner for C# yet
Read the code here, then run it in your own toolchain or a ready-made cloud workspace.
Expected output
{
"pooledA": 0.8272727272727273,
"pooledB": 0.325,
"easy": [
0.9,
0.95
],
"hard": [
0.1,
0.2
]
}Prefer your own machine? Every file is in the course repository · open it in Codespaces.
Lesson notes
The rule
association ≠ effect of intervention