Find agent skills by outcome

- `nstart > 1` recommended (e.g.

`nstart = 25`): runs algorithm from multiple random starts

- `prcomp` uses SVD (numerically superior); `princomp` uses `eigen` on covariance (less stable

N-1 vs N scaling).

- `method = euclidean` (default). Other options: `manhattan`

`maximum`

- Methods: `holm` (default)

`BH` (Benjamini-Hochberg FDR)

- Paired test: `paired = TRUE` uses signed-rank test (Wilcoxon)

not rank-sum (Mann-Whitney).

- For tables larger than 2x2

uses simulation (`simulate.p.value = TRUE`) or network algorithm.

- `exact = TRUE` by default for small samples with no ties. With ties

normal approximation used.

- `simulate.p.value = TRUE`: Monte Carlo with `B = 2000` replicates (min p ~ 0.0005). Simulation assumes fixed marginals (Fisher-style sampling

not the chi-sq assumption).

- Return object includes `$expected`

`$residuals` (Pearson)

- For goodness-of-fit: pass a vector

not a matrix. `p` must sum to 1 (or set `rescale.p = TRUE`).

> Non-obvious behaviors

gotchas

- `anova(model1

model2)`: F-test comparing nested models.

- `uniroot`: finds a root of `f` in `[lower

upper]`. **Requires** `f(lower)` and `f(upper)` to have opposite signs.

- Returns object with `predict`

`print`

Coding & Debugging

- `SANN` (simulated annealing): convergence code is always 0 — it never fails. `maxit` = total function evals (default 10000)

no other stopping criterion.

- Default method is Nelder-Mead (no gradients

robust but slow). Poor for 1D — use `Brent` or `optimize()`.

- `lowess` parameter is `f` (span

default 2/3); `loess` parameter is `span` (default 0.75).

- `predict.glm` with `type = response` gives predictions on the response scale (e.g.

probabilities for logistic); `type = link` (default) gives on the link scale.

- `lowess` is the older function; returns `list(x

y)` — cannot predict at new points.

- Direction: `direction = both` (default)

`forward`

Creative

- Self-starting models (`SSlogis`

`SSasymp`

- For Type III SS

use `car::Anova()` or set contrasts to `contr.sum`/`contr.helmert`.

Creative

- `aov` is a wrapper around `lm` that stores extra info for balanced ANOVA. For unbalanced designs

Type I SS (sequential) are computed — order of terms matters.

- Quasi-families (`quasibinomial`

`quasipoisson`) allow overdispersion — no AIC is computed.