Chapter 28 Appendix: Notation and Function Reference
This appendix consolidates the notation used across the book and maps each object to its primary R implementation pattern. It is intended as a quick lookup for readers moving between theoretical sections and code-first workflows.
Unless otherwise noted, all functions in this appendix come from the NNS package. In executable code, load the package once and then call functions directly as shown in the table entries below.
The crosswalk gives the intended Python counterpart names in the nns package (distributed as ovvo-nns; see the Preface for conventions). R is the canonical implementation for this edition. New arguments, output labels, and numerical repairs must be checked against the installed Python version before assuming implementation parity:
28.1 R–Python Function Crosswalk
The R name is the canonical reference used throughout the book; the Python name follows the snake_case convention.
| R function | Python function | Purpose |
|---|---|---|
LPM / UPM |
nns.lpm / nns.upm |
Lower / upper partial moments |
LPM.ratio / UPM.ratio |
nns.lpm_ratio / nns.upm_ratio |
Frequency probabilities at degree zero; directional area shares at positive degrees |
LPM.VaR / UPM.VaR |
nns.lpm_var / nns.upm_var |
Lower / upper tail quantile inversion |
Co.LPM / Co.UPM |
nns.co_lpm / nns.co_upm |
Concordant co-partial moments |
D.LPM / D.UPM |
nns.d_lpm / nns.d_upm |
Divergent co-partial moments |
PM.matrix |
nns.pm_matrix |
Partial-moment covariance decomposition |
NNS.dep |
nns.nns_dep |
Nonlinear dependence measure |
NNS.copula |
nns.nns_copula |
Multivariate dependence / copula view |
NNS.caus |
nns.nns_causation |
Directional causation diagnostic |
NNS.CDF |
nns.nns_cdf |
Partial-moment CDF estimation |
NNS.ANOVA |
nns.nns_anova |
Partial-moment ANOVA certainty |
NNS.SS |
nns.nns_ss |
Stochastic superiority |
NNS.FSD / NNS.SSD / NNS.TSD |
nns.fsd / nns.ssd / nns.tsd |
Stochastic dominance tests |
NNS.part |
nns.nns_part |
Recursive mean-split partitioning |
NNS.reg |
nns.nns_reg |
Nonparametric regression |
NNS.boost |
nns.nns_boost |
Feature-screening ensemble learner |
NNS.stack |
nns.nns_stack |
Optimized stacked ensemble |
NNS.seas |
nns.nns_seas |
Seasonality detection |
NNS.ARMA |
nns.nns_arma |
Nonlinear autoregressive forecasting |
NNS.ARMA.optim |
nns.nns_arma_optim |
Seasonal-factor optimization |
NNS.VAR |
nns.nns_var |
Nonparametric vector autoregression |
NNS.meboot |
nns.nns_meboot |
Maximum-entropy bootstrap |
NNS.MC |
nns.nns_mc |
Meboot-based Monte Carlo wrapper |
NNS.mode |
nns.nns_mode |
Mode estimation |
NNS.moments |
nns.nns_moments |
Partial-moment classical moments |
NNS.diff / dy.dx |
nns.nns_diff / nns.dy_dx |
Numerical differentiation |
NNS.norm / NNS.rescale |
nns.nns_norm / nns.nns_rescale |
Normalization and rescaling |
28.2 Core directional operators and partial moments
| Symbol | Definition | Interpretation | R function / pattern |
|---|---|---|---|
| \(x^+\) | \(\max(x,0)\) | Positive-part operator | pmax(x, 0) |
| \((X-t)^+\) | \(\max(X-t,0)\) | Deviation above benchmark \(t\) | internal to UPM(...) |
| \((t-X)^+\) | \(\max(t-X,0)\) | Deviation below benchmark \(t\) | internal to LPM(...) |
| \(L_r(t;X)\) | \(E[(t-X)_+^r]\) | Lower partial moment, degree \(r\) | LPM(r, t, X) |
| \(U_r(t;X)\) | \(E[(X-t)_+^r]\) | Upper partial moment, degree \(r\) | UPM(r, t, X) |
| \(L_r/(L_r+U_r)\) | Degree-\(r\) lower ratio | Frequency below \(t\) at degree zero; lower severity share otherwise | LPM.ratio(r, t, X) |
| \(U_r/(L_r+U_r)\) | Degree-\(r\) upper ratio | Frequency above \(t\) at degree zero; upper severity share otherwise | UPM.ratio(r, t, X) |
28.3 Co-partial moments, dependence, and causation
| Symbol | Definition / role | R function |
|---|---|---|
| \(CoLPM, CoUPM\) | Concordant lower/upper co-partial moments | Co.LPM(...), Co.UPM(...) |
| \(DLPM, DUPM\) | Divergent lower/upper co-partial moments | D.LPM(...), D.UPM(...) |
| \(NNS.dep(X,Y)\) | Global nonlinear dependence measure | NNS.dep(x, y) |
| \(NNS.copula(X)\) | Nonparametric dependence geometry / copula view | NNS.copula(cbind(x, y)) |
| \(NNS.caus(X,Y)\) | Directional causation diagnostic | NNS.caus(x, y) |
28.4 Distribution comparison, dominance, and interval objects
| Symbol | Definition / role | R function |
|---|---|---|
| \(F^{(0)}(t)\) | Degree-zero empirical CDF (step measure) | ecdf(x)(t) or LPM.ratio(0, t, x) |
| \(F^{(1)}(t)\) | Degree-one area ratio: the share of deviation area below \(t\), not \(P(X \le t)\) | LPM.ratio(1, t, x) |
| \(p = P(X' > Y')\) | Directional exceedance probability for pairwise comparison | estimated by cross-sample indicator averages |
| \(\text{Certainty}_{\text{ANOVA}}\) | Descriptive target agreement with a finite-sample adjustment; not a p-value or power | NNS.ANOVA(...) |
| FSD / SSD / TSD | Ordinary dominance at the stated orders; the legacy continuous FSD option instead uses Omega/expectile order | NNS.FSD(...), NNS.SSD(...), NNS.TSD(...) |
| \(Q^-_{d}(\alpha)\) | Lower degree-\(d\) quantile | LPM.VaR(alpha, degree = d, x) |
| \(Q^+_{d}(\alpha)\) | Upper degree-\(d\) quantile | UPM.VaR(alpha, degree = d, x) |
| Directional band | \([Q^-_d(\alpha/2),Q^+_d(\alpha/2)]\): frequency target at degree zero, severity target otherwise; future coverage needs validation | LPM.VaR(...) + UPM.VaR(...) |
LPM.VaR(percentile, degree, x)
Lower-tail threshold operator obtained by inverting the degree-specific lower partial-moment representation.
Interpretation by degree:
degree = 0: type-7 interpolated empirical lower quantile,degree = 1: severity-weighted lower threshold based on directional magnitude,degree = 2: extreme-deviation-sensitive lower threshold. In finance, the degree-zero case is commonly called VaR, but the operator is more general than that label.
UPM.VaR(percentile, degree, x)
Upper-tail analog of LPM.VaR, used for right-tail threshold selection and interval construction.
28.5 Directional Decision Regions Crosswalk (Classical → NNS)
To maintain continuity with Chapter 23’s directional decision-region framing, the table below maps common classical statistics and procedures to their directional NNS counterparts.
| Classical statistic / workflow | Typical classical role | Directional NNS counterpart | Reference chapter | Notes |
|---|---|---|---|---|
| Pearson correlation | Linear association summary | NNS.dep(x, y) |
Chapter 10 | Captures nonlinear and asymmetric dependence, not only linear co-movement. |
| Parametric VaR / empirical quantile VaR | Tail-loss thresholding | LPM.VaR(alpha, degree, x) (degree-dependent) |
Chapter 16 | Degree controls sensitivity to tail severity beyond degree-0 quantiles. |
| Upper-tail quantile threshold | Right-tail risk/opportunity cutoff | UPM.VaR(alpha, degree, x) (degree-dependent) |
Chapter 16 | Upper-tail analog to LPM.VaR for asymmetric interval construction. |
| Classical ANOVA (mean-comparison test) | Group-level location comparison | NNS.ANOVA(...) (degree-dependent CDF benchmarking) |
Chapter 15 | Agreement certainty is benchmarked through directional CDF-style deviations. |
| Linear Granger-style directional inference | Lead-lag direction under linear structure | NNS.caus(x, y) |
Chapter 14 | Measures directional lagged association; does not by itself identify a structural cause or condition on own-history dynamics. |
| Copula / Joint Tail Dependency | Joint probability of concurrent outcomes | Co.LPM(degree_lpm, x, y, target_x, target_y) / Co.UPM(degree_upm, x, y, target_x, target_y) |
Chapters 10 and 13 | Degree zero counts joint events; positive degrees measure joint directional magnitudes, with the requisite finite moments. |
| Mean-variance interval heuristics | Uncertainty bands under Gaussian assumptions | LPM.VaR(...) + UPM.VaR(...) (degree-dependent bounds) |
Chapter 17 | Constructs directional thresholds without selecting a normal family; coverage requires a separate sampling and calibration argument. |
28.6 Regression and forecasting workflow objects
| Symbol | Definition / role | R function | Reference chapter |
|---|---|---|---|
| \(\hat m_{NNS}(x)\) | NNS fitted location; the mean reducer targets conditional means under the stated conditions | NNS.reg(x, y) |
Chapter 22 |
| Global residual distribution | Pooled in-sample residuals, not a local conditional error law | via NNS.reg(...)$Fitted.xy$residuals outputs |
Chapter 22 |
| Residual band at \(x_0\) | Fitted location plus global residual-quantile offsets | NNS.reg(..., point.est = x0, confidence.interval = ...) |
Chapter 17 |
| Regime-specific directional dependence | Time-local / state-local dependence | NNS.dep(...) on rolling/segmented windows |
Chapter 10 |
28.7 A.3 Technical Note: Adaptive Order and Consistency Conditions
Chapter 18 gives a conditional consistency argument for cell means under an appropriate sampling and partition design. Its assumptions include shrinking cell diameters, diverging local occupancy, control of conditional variation, and appropriate locality of any subsequent interpolation or averaging. Response-adaptive partition selection requires its own conditions; a proof for predictor-only partitions cannot simply be reused after selection on outcomes.
The current default order heuristic is bounded. A fixed minimum cell size does not establish diverging occupancy, and a bound on depth does not establish shrinking diameter as the sample grows. The default implementation therefore does not inherit a universal consistency theorem or an oracle convergence rate from the abstract construction. Such claims require a specified sequence of depths/neighbour counts and a proof that the actual estimator satisfies the conditions. Other location reducers, including gravity, need analysis for their own targets rather than an automatic conditional-mean claim.
28.8 Validation and compatibility choices
| Function | Current choice | Interpretation |
|---|---|---|
NNS.reg, NNS.stack |
type = "REG" |
Explicit numeric regression, including two-valued responses; predictions are not restricted to class labels. |
NNS.ARMA.optim |
bias.correction = "none" (default) or "split" |
No shift by default; split estimates the shift earlier and accepts it later within internal validation. Outer evaluation is still needed. |
NNS.VAR |
validation = "pipeline" or "observed" |
Pipeline predicts the held-out feature rows from the earlier origin; observed is the conditional-regression compatibility mode. |
NNS.seas |
dispersion = "auto" or "absolute" |
Auto preserves the historical sign-based CV/Gini choice; absolute supports translation-invariant interval-scale comparisons. |
NNS.meboot |
expand.sd and expand.residual.sd |
Level and residual expansion are separate; force.clt uses an iid mean-scale reference. |
NNS.FSD |
type = "continuous" |
Numerical area order is retained and returns OMEGA result labels. |
Use the installed package help for complete signatures. The table names statistical choices; it does not replace the assumptions and qualifications in the corresponding chapters.