Find agent skills by outcome

# write.csv(df

%s_clean.csv

fit <- lm(%s ~ %s

data = df)

# fit_aov <- aov(%s ~ %s

data = df)

# t.test(%s ~ %s

data = df)

if (is.null(data_file)) data_file <- paste0(project_name

.csv)

check_data <- function(df

top_n_levels = 8) {

# saveRDS(df

data_clean.rds)

Writing & Content

# write.csv(df

data_clean.csv

# fit_aov <- aov(outcome_var ~ group_var

data = df)

t.test(outcome_var ~ group_var

data = df)

abline(lm(outcome_var ~ predictor

data = df)

df <- read.csv(your_data.csv

stringsAsFactors = FALSE)

Coding & Debugging

- `arrows`: `code = 1` (head at start)

`code = 2` (head at end

- Formula interface: `pairs(~ var1 + var2 + var3

data = df)`.

- Formula interface: `cor.test(~ x + y

data = df)` — note the `~` with no LHS.

- Default kernel: `gaussian`. Range of density extends beyond data range (controlled by `cut`

default 3 bandwidths).

- Default bandwidth: `bw = nrd0` (Silverman's rule of thumb). For multimodal data

consider `SJ` or `bcv`.

Creative

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

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

- Returns a matrix of p-values

not test statistics.

- For composite hypotheses (parameters estimated from data)

p-values are **conservative** (too large). Use `dgof` or `ks.test` with `exact = NULL` for discrete distributions.

Coding & Debugging

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

no other stopping criterion.

- If data fits too exactly (no residual noise)

convergence check fails — use `control = list(scaleOffset = 1)` or jitter data.

Creative

- Self-starting models (`SSlogis`

`SSasymp`

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.