General
PromptBeginner5 minmarkdown
> Non-obvious behaviors
gotchas
0
Explore
Find agent skills by outcome
118,479 skills indexed with the new KISS metadata standard.
Showing 24 of 118,479Categories: Data, General, Coding & Debugging
Coding & Debugging
PromptBeginner5 minmarkdown
- `SANN` (simulated annealing): convergence code is **always 0** — it never fails. `maxit` = total function evals (default 10000)
no other stopping criterion.
0
General
PromptBeginner5 minmarkdown
- `uniroot`: finds a root of `f` in `[lower
upper]`. **Requires** `f(lower)` and `f(upper)` to have opposite signs.
0
General
PromptBeginner5 minmarkdown
- `anova(model1
model2)`: F-test comparing nested models.
0
General
PromptBeginner5 minmarkdown
- Returns object with `predict`
`print`
0
General
PromptBeginner5 minmarkdown
- Default method is Nelder-Mead (no gradients
robust but slow). Poor for 1D — use `Brent` or `optimize()`.
0
General
PromptBeginner5 minmarkdown
- `lowess` parameter is `f` (span
default 2/3); `loess` parameter is `span` (default 0.75).
0
General
PromptBeginner5 minmarkdown
- Direction: `direction = both` (default)
`forward`
0
General
PromptBeginner5 minmarkdown
- `lowess` is the older function; returns `list(x
y)` — cannot predict at new points.
0
General
PromptBeginner5 minmarkdown
- `predict.glm` with `type = response` gives predictions on the response scale (e.g.
probabilities for logistic); `type = link` (default) gives on the link scale.
0
Data
PromptBeginner5 minmarkdown
- If data fits too exactly (no residual noise)
convergence check fails — use `control = list(scaleOffset = 1)` or jitter data.
0
General
PromptBeginner5 minmarkdown
- For Type III SS
use `car::Anova()` or set contrasts to `contr.sum`/`contr.helmert`.
0
General
PromptBeginner5 minmarkdown
- `anova(glm_obj
test = Chisq)` for deviance-based tests; `F` is invalid for non-Gaussian families.
0
General
PromptBeginner5 minmarkdown
- Quasi-families (`quasibinomial`
`quasipoisson`) allow overdispersion — no AIC is computed.
0
General
PromptBeginner5 minmarkdown
- `weights` in `glm` means **prior weights** (not frequency weights) — for frequency weights
use the cbind trick or offset.
0
General
PromptBeginner5 minmarkdown
- `binomial` accepts response as: 0/1 vector
logical
0
General
PromptBeginner5 minmarkdown
- Common families: `binomial(link = logit)`
`poisson(link = log)`
0
General
PromptBeginner5 minmarkdown
- Column names from `model.matrix` can be surprising: e.g.
`factorLevelName` concatenation.
0
General
PromptBeginner5 minmarkdown
- `terms` object attributes: `order` (interaction order per term)
`intercept`
0
General
PromptBeginner5 minmarkdown
- Default `family = gaussian(link = identity)` — `glm()` with no `family` silently fits OLS (same as `lm`
but slower and with deviance-based output).
0
General
PromptBeginner5 minmarkdown
- `I()` is required to use arithmetic operators literally: `y ~ x + I(x^2)`. Without `I()`
`^` means interaction crossing.
0
General
PromptBeginner5 minmarkdown
> Non-obvious behaviors
gotchas
0
General
PromptBeginner5 minmarkdown
- Captures output from `cat`
`print`
0
General
PromptBeginner5 minmarkdown
- `as.is = TRUE` (recommended): keeps characters as character
not factor.
0