Linear log-odds score before logistic conversion

gh-eq-48b8db92e4390604source-continuation-equation

fixed champion, a fixed feature vector, and no DB. It computes log odds = intercept + SUM(coefficient_j * value_j) in shopspring/decimal, applies the logistic (a single isolated float conversion, re quantized to

Read the equation

The model adds an intercept to the sum of each coefficient times its corresponding feature. Applying a logistic transform converts that log-odds score into a value between zero and one. The source uses decimal arithmetic for the linear combination and an isolated floating-point conversion for the nonlinear transform; actual predictive calibration depends on training and evaluation, not only this formula.

Terms & definitions

Term	Meaning in this source
intercept	Constant model offset
coefficient_j	Weight of feature j
value_j	Feature value
log_odds	Linear model output

How it works

Read the definition from its right-hand expression to the quantity on the left. Its surrounding source records the intended input, state, or parameter context. Where the source leaves an operation unspecified, this page preserves that gap.

Assumptions & limits

Source recovery establishes that the expression appears in this record. Its mathematical validity, implementation behavior, and interpretation must be assessed separately. See the linked source and research notes for the checks available for this version.

Original context

Read the surrounding source passage

// ExecuteIntent path) and extracts the declared scalar feature from its result. The extractor closes over the
// bound args (the segment keys). It is the FIXED, reviewed set of features a MAG intent may read; a feature
// intent with no extractor is refused, so a model can never read a feature the platform did not author an
// extractor for.
type featureExtractor func(ctx context.Context, db *tenant.DB, args []any) (decimal.Decimal, string, error)

// scoreModelIntent is the pure serving math, separated from the database read so it is unit testable with a
// fixed champion, a fixed feature vector, and no DB. It computes log odds = intercept + SUM(coefficient_j *
// value_j) in shopspring/decimal, applies the logistic (a single isolated float conversion, re quantized to
// decimal immediately), applies the calibration map, and returns the calibrated probability plus the ordered
// contribution trace. Every SURFACED number (the value, the coefficient, the contribution, the probability) is
// decimal; the only float is the transient sigmoid argument, which never surfaces.
func scoreModelIntent(model championModel, feats []Contribution) (decimal.Decimal, []Contribution) {
	// The additive log odds in decimal: the intercept plus every feature's coefficient times value. Each
	// feature carries its own contribution (coefficient times value) as the citation.
	logOdds := model.Intercept
	trace := make([]Contribution, 0, len(feats)+1)
	trace = append(trace, Contribution{

Exact original notation

fixed champion, a fixed feature vector, and no DB. It computes log odds = intercept + SUM(coefficient_j *
value_j) in shopspring/decimal, applies the logistic (a single isolated float conversion, re quantized to