relevances.sleela
churn/1/relevances.sleela · 109 line(s) · sleela
#sleela 2.1
// churn/1/relevances.sleela -- "Longs": relevances among US economic length series.
//
// Each `Long` is a length series (a magnitude that runs over 2019..2024). The
// model computes, for every pair of lengths, the Pearson correlation r via the
// native `inference` library and classifies the relevance the way Nordshrift
// classifies object relevance -- Direct / Model / None -- by |r|:
// |r| >= 0.85 -> Direct (co-occupy the same material extent)
// |r| >= 0.50 -> Model (looser, modeled association)
// else -> None (no material co-relation over the window)
// It prints a mathematical relevance matrix and a paragraph reading.
//
// Data are real, dated, sourced annual figures (see longs.csv / README.md).
// A relevance is a measured association over the window; ASSOCIATION != CAUSATION.
import math;
import inference;
struct Long {
String name;
String symbol;
// six yearly readings 2019..2024
double v1; double v2; double v3; double v4; double v5; double v6;
}
class Longs {
// Pearson r between two Longs, folding their six readings into the
// correlation running sums (X = long a, Y = long b).
double relevanceR(Long a, Long b) {
double n = 6.0;
double sx = a.v1 + a.v2 + a.v3 + a.v4 + a.v5 + a.v6;
double sy = b.v1 + b.v2 + b.v3 + b.v4 + b.v5 + b.v6;
double sxx = a.v1*a.v1 + a.v2*a.v2 + a.v3*a.v3 + a.v4*a.v4 + a.v5*a.v5 + a.v6*a.v6;
double syy = b.v1*b.v1 + b.v2*b.v2 + b.v3*b.v3 + b.v4*b.v4 + b.v5*b.v5 + b.v6*b.v6;
double sxy = a.v1*b.v1 + a.v2*b.v2 + a.v3*b.v3 + a.v4*b.v4 + a.v5*b.v5 + a.v6*b.v6;
return inference.correlation(n, sx, sy, sxx, syy, sxy);
}
// Class label from |r|, mirroring Nordshrift's Direct/Model/None relevance.
String relevanceClass(double r) {
double ar = math.abs(r);
if (ar >= 0.85) { return "Direct"; }
if (ar >= 0.5) { return "Model"; }
return "None";
}
// Direction word from the sign of r.
String direction(double r) {
if (r > 0.0) { return "co-extending"; }
if (r < 0.0) { return "opposing"; }
return "unrelated";
}
// One matrix row + one paragraph line for a pair.
void reportPair(Long a, Long b) {
double r = relevanceR(a, b);
String cls = relevanceClass(r);
print(" " + a.symbol + " ~ " + b.symbol + " : r=" + r + " [" + cls + "]");
}
void sentence(Long a, Long b) {
double r = relevanceR(a, b);
String cls = relevanceClass(r);
print(" " + a.name + " and " + b.name + " hold a " + cls
+ " relevance (r=" + r + ", " + direction(r)
+ "): they occupy the same material extent "
+ occupancy(cls) + ".");
}
String occupancy(String cls) {
if (cls == "Direct") { return "directly over 2019-2024"; }
if (cls == "Model") { return "only through a looser modeled association"; }
return "with no material co-relation evidenced";
}
void main() {
// --- the lengths (real sourced annual figures; see longs.csv) ---
Long gdp = new Long();
gdp.name = "Nominal GDP"; gdp.symbol = "GDP";
gdp.v1 = 21.5; gdp.v2 = 21.4; gdp.v3 = 23.7; gdp.v4 = 26.0; gdp.v5 = 27.7; gdp.v6 = 29.2;
Long debt = new Long();
debt.name = "Total federal debt"; debt.symbol = "DEBT";
debt.v1 = 22.7; debt.v2 = 26.9; debt.v3 = 28.4; debt.v4 = 30.9; debt.v5 = 33.2; debt.v6 = 35.5;
Long cpi = new Long();
cpi.name = "Consumer price level"; cpi.symbol = "CPI";
cpi.v1 = 255.7; cpi.v2 = 258.8; cpi.v3 = 271.0; cpi.v4 = 292.7; cpi.v5 = 304.7; cpi.v6 = 313.7;
print("=== /1 Longs -- relevances among US economic lengths (2019-2024) ===");
print("Relevance class by |r|: Direct >= 0.85, Model >= 0.50, else None.");
print("");
print("[mathematical] relevance matrix");
reportPair(gdp, debt);
reportPair(gdp, cpi);
reportPair(debt, cpi);
print("");
print("[evaluation] how the lengths co-relate and occupy materially");
sentence(gdp, debt);
sentence(gdp, cpi);
sentence(debt, cpi);
print("");
print("Note: real sourced data (BEA / US Treasury+GAO / BLS); a relevance");
print("is a measured association over 2019-2024, not a causal claim.");
}
}