Package 'EnTraineR'

Convert an EntraineR response to character

Description

Convert an EntraineR response to character

Usage

## S3 method for class 'entrainer_response'
as.character(x, ...)

Arguments

x	Object returned by gemini_generate().
...	Unused.

Value

Generated text.

---

River deforestation: air and water temperatures before/after

Description

Monitoring data of water and air temperatures before and after riparian deforestation. Useful to illustrate linear regression with an interaction (Temp_air * Deforestation).

Usage

data(deforestation)

Format

A data frame with 56 rows and 3 variables:

Temp_water

numeric; water temperature (deg C).

Temp_air

numeric; air temperature (deg C).

Deforestation

factor with 2 levels: "BEFORE", "AFTER". 28 periods each.

Details

Brief summary (indicative): Temp_water min ~ 0.55, median ~ 9.28, max ~ 18.89; Temp_air min ~ -3.04, median ~ 6.53, max ~ 15.75.

Examples

data(deforestation)
str(deforestation)
table(deforestation$Deforestation)


# Linear model with interaction (FactoMineR):
fit <- FactoMineR::LinearModel(
  Temp_water ~ Temp_air * Deforestation,
  data = deforestation,
  selection = "none"
)
print(fit)

---

Extract the response from an EntraineR prompt/result object

Description

Extract the response from an EntraineR prompt/result object

Usage

entrainer_response(x)

Arguments

x	Object returned by an EntraineR trainer.

Value

The stored LLM response, or NULL.

---

Generate text with Google Gemini (Generative Language API) - robust w/ retries

Description

Minimal wrapper around the Generative Language API ':generateContent' endpoint for text prompts, with retries, exponential backoff, clearer errors, and optional output compilation (HTML/DOCX). Files are opened only when 'open = TRUE'.

Usage

gemini_generate(
  prompt,
  model = "gemini-2.5-flash",
  api_key = Sys.getenv("GEMINI_API_KEY"),
  user_agent = NULL,
  base_url = "https://generativelanguage.googleapis.com/v1beta",
  temperature = NULL,
  top_p = NULL,
  top_k = NULL,
  max_output_tokens = NULL,
  stop_sequences = NULL,
  system_instruction = NULL,
  safety_settings = NULL,
  seed = NULL,
  timeout = 120,
  verbose = FALSE,
  max_tries = 5,
  backoff_base = 0.8,
  backoff_cap = 8,
  force_markdown = TRUE,
  compile_to = c("none", "html", "docx"),
  output_path = NULL,
  open = interactive()
)

Arguments

prompt	Character scalar. The user prompt (plain text).
model	Character scalar. Gemini model id (e.g., "gemini-2.5-flash", "gemini-2.5-pro"). You may also pass "models/..." and it will be normalized.
api_key	Character scalar. API key. Defaults to env var 'GEMINI_API_KEY'.
user_agent	Character scalar. If NULL, a dynamic value is used.
base_url	Character scalar. API base URL.
temperature	Optional numeric in [0, 2].
top_p	Optional numeric in (0, 1].
top_k	Optional integer >= 1.
max_output_tokens	Optional integer > 0.
stop_sequences	Optional character vector.
system_instruction	Optional character scalar.
safety_settings	Optional list passed as-is to the API.
seed	Optional integer seed.
timeout	Numeric seconds for request timeout (default 120).
verbose	Logical; if TRUE, prints URL/retries.
max_tries	Integer. Max attempts (default 5).
backoff_base	Numeric. Initial backoff seconds (default 0.8).
backoff_cap	Numeric. Max backoff seconds (default 8).
force_markdown	Logical. If TRUE, instructs the model to answer in Markdown.
compile_to	Character scalar. One of c("none","html","docx").
output_path	Optional character scalar. Destination file for HTML/DOCX output. If NULL, a temporary file is created.
open	Logical; if TRUE, open the generated HTML/DOCX file. Defaults to 'interactive()'.

Value

An object of class 'entrainer_response' with a stable structure. The generated text is available in '$text'/'$markdown'; 'html_path' or 'docx_path' are populated when 'compile_to' is '"html"' or '"docx"'.

Privacy

This function sends 'prompt' to the Google Generative Language API. Do not include confidential data unless this is intended and allowed in your context.

---

Ham: sensory descriptors and overall liking

Description

Sensory profile of hams (quantitative attributes) and an overall liking score. Useful to illustrate multiple regression and the joint reading of per-term F tests and coefficient T tests.

Usage

data(ham)

Format

A data frame with 21 rows (hams) and 15 variables:

Juiciness

numeric

Crispy

numeric

Tenderness

numeric

Pasty

numeric

Fibrous

numeric

Salty

numeric

Sweet

numeric

Meaty

numeric

Seasoned

numeric

Metallic

numeric

Ammoniated

numeric

Fatty

numeric

Braised

numeric

Lactic

numeric

Overall liking

numeric; overall acceptability score

Details

Brief summary (indicative): median Juiciness ~ 3.0; median Tenderness ~ 6.0; mean Salty ~ 5.52; median Overall liking ~ 6.5.

Examples

data(ham)
summary(ham)


# Multiple regression without selection (FactoMineR):
fit <- FactoMineR::LinearModel(
  `Overall liking` ~ .,
  data = ham,
  selection = "none"
)
print(fit)

---

Poussin: weight by brooding temperature and sex

Description

Chick weights measured under three brooding temperatures, with sex recorded. Useful for ANOVA and linear models with categorical factors.

Usage

data(poussin)

Format

A data frame with 45 rows and 3 variables:

Temperature

factor with 3 levels: "T1", "T2", "T3" (15 each).

Gender

factor with 2 levels: "Female", "Male" (about 20 and 25).

Weight

numeric; weight (units as provided).

Details

Brief summary (indicative): Weight min ~ 15, median ~ 23, max ~ 33.

Examples

data(poussin)
with(poussin, table(Temperature, Gender))
boxplot(Weight ~ Temperature, data = poussin,
        main = "Poussin weight by temperature")
# Two-factor ANOVA (base stats):
fit <- stats::aov(Weight ~ Temperature * Gender, data = poussin)
summary(fit)

---

Backward-compatible print method name

Description

Backward-compatible print method name

Usage

## S3 method for class 'entrainer_llm_result'
print(x, ...)

Arguments

x	Object returned by older EntraineR generators.
...	Unused.

Value

Invisibly returns x.

---

Print an EntraineR prompt/result compactly

Description

Print an EntraineR prompt/result compactly

Usage

## S3 method for class 'entrainer_prompt'
print(x, ...)

Arguments

x	Object returned by an EntraineR trainer.
...	Unused.

Value

Invisibly returns x.

---

Print an EntraineR response compactly

Description

Print an EntraineR response compactly

Usage

## S3 method for class 'entrainer_response'
print(x, ...)

Arguments

x	Object returned by gemini_generate().
...	Unused.

Value

Invisibly returns x.

---

Trainer: Interpret ANOVA (AovSum) with an LLM-ready prompt

Description

Builds an English-only, audience-tailored prompt to interpret an ANOVA produced by FactoMineR::AovSum. The function never invents numbers: it only passes verbatim excerpts to the LLM and instructs how to interpret deviations (sum-to-zero coding) as performance drivers.

Usage

trainer_aovsum(
  x,
  introduction = NULL,
  alpha = 0.05,
  t_test = NULL,
  audience = c("beginner", "applied", "advanced"),
  summary_only = FALSE,
  llm_model = "llama3",
  generate = FALSE,
  llm_engine = c("ollama", "gemini", "none"),
  ...
)

trainer_AovSum(aovsum_obj, ...)

Arguments

x	An object whose printed output contains sections named "Ftest" and "Ttest" (e.g., FactoMineR::AovSum()).
introduction	Optional character context paragraph for the analysis. Defaults to a generic description.
alpha	Numeric significance level used as an instruction for the LLM. Default 0.05.
t_test	Optional character vector to filter the T-test section by factor names and/or interactions (e.g. "Factor A" or "Factor A:Factor B").
audience	Target audience, one of c("beginner","applied","advanced").
summary_only	Logical; if TRUE, return a compact 3-bullet executive summary.
llm_model	Character model name for the generator (e.g., "llama3").
generate	Logical; if TRUE, calls trainer_core_generate_or_return().
llm_engine	Character; backend engine: "ollama", "gemini", or "none".
...	Passed to the selected LLM backend when 'generate = TRUE'.
aovsum_obj	Deprecated name for 'x'.

Value

An 'entrainer_prompt' object. It behaves like a character string for 'cat()'/printing and stores LLM metadata and response as attributes when 'generate = TRUE'.

Privacy

If 'generate = TRUE' and 'llm_engine' is not '"none"', the prompt is sent to the selected LLM backend. With external providers such as Gemini, this may include excerpts of statistical outputs and user-provided context.

Examples

## Not run: 
# Example 1: SensoMineR chocolates (requires SensoMineR)
if (requireNamespace("SensoMineR", quietly = TRUE)) {
# Load data from SensoMineR
data("chocolates", package = "SensoMineR")
# ANOVA summary with Product and Panelist
res <- FactoMineR::AovSum(Granular ~ Product * Panelist, data = sensochoc)

intro <- "Six chocolates have been evaluated by a sensory panel,
during two days, according to a sensory attribute: granular.
The panel has been trained according to this attribute
and panellists should be reproducible when rating this attribute."
intro <- gsub("\n", " ", intro)
intro <- gsub("\\s+", " ", intro)
cat(intro)

prompt <- trainer_aovsum(res, audience = "beginner",
                 t_test = c("Product", "Panelist"),
                 introduction = intro)
cat(prompt)

res <- gemini_generate(prompt, compile_to = "html")
}

# Example 2: Poussin dataset (shipped with this package)
data(poussin)
intro <- "For incubation, 45 chicken eggs were randomly assigned to three batches of 15.
Three treatments (different incubation temperatures) were then applied to the batches.
We assume that after hatching, all chicks were raised under identical conditions
and then weighed at a standard reference age.
At that time, the sex of the chicks - a factor known beforehand to cause
significant weight differences - could also be observed.
The objective is to choose the treatment that maximizes chick weight."
intro <- gsub("\n", " ", intro)
intro <- gsub("\\s+", " ", intro)
cat(intro)

res <- FactoMineR::AovSum(Weight ~ Gender * Temperature, data = poussin)

prompt <- trainer_aovsum(res,
                 audience = "beginner",
                 t_test = c("Gender", "Temperature"),
                 introduction = intro)
cat(prompt)

res <- gemini_generate(prompt, compile_to = "html")

## End(Not run)

---

Interpret a chi-squared test (chisq.test) with an audience-aware LLM prompt

Description

Builds a clear, audience-tailored prompt to interpret base R stats::chisq.test() results, handling both goodness-of-fit and contingency-table tests. Aligned with other EntraineR trainers: no invented numbers; audience-specific guidance.

Usage

trainer_chisq_test(
  csq_obj,
  introduction = NULL,
  alpha = 0.05,
  audience = c("beginner", "applied", "advanced"),
  summary_only = FALSE,
  llm_model = "llama3",
  generate = FALSE,
  llm_engine = c("ollama", "gemini", "none"),
  ...
)

Arguments

csq_obj	An htest object returned by stats::chisq.test().
introduction	Optional character string giving the study context.
alpha	Numeric significance level (default 0.05).
audience	One of c("beginner","applied","advanced").
summary_only	Logical; if TRUE, return a 3-bullet executive summary regardless of audience depth (uses trainer_core_summary_only_block()).
llm_model	Character; model name for the generator (default "llama3").
generate	Logical; if TRUE, call the generator and return prompt + response.
llm_engine	Character; backend engine: "ollama", "gemini", or "none".
...	Passed to the selected LLM backend when 'generate = TRUE'.

Value

An 'entrainer_prompt' object. It behaves like a character string for 'cat()'/printing and stores LLM metadata and response as attributes when 'generate = TRUE'.

Privacy

If 'generate = TRUE' and 'llm_engine' is not '"none"', the prompt is sent to the selected LLM backend. With external providers such as Gemini, this may include excerpts of statistical outputs and user-provided context.

Examples

# GOF
set.seed(1); x <- c(18, 22, 20, 25, 15)
csq1 <- chisq.test(x, p = rep(1/5, 5))
cat(trainer_chisq_test(csq1, audience = "beginner"))

# Contingency (independence)
tbl <- matrix(c(12,5,7,9), nrow=2)
csq2 <- chisq.test(tbl)  # Yates for 2x2 by default
cat(trainer_chisq_test(csq2, audience = "applied"))

---

Interpret a correlation test (cor.test) with an audience-aware LLM prompt

Description

Builds a clear, audience-tailored prompt to interpret stats::cor.test() results for Pearson, Spearman, or Kendall correlation. Supports three audiences ("beginner", "applied", "advanced") and an optional summary_only mode.

Usage

trainer_cor_test(
  ct_obj,
  introduction = NULL,
  alpha = 0.05,
  audience = c("beginner", "applied", "advanced"),
  summary_only = FALSE,
  llm_model = "llama3",
  generate = FALSE,
  llm_engine = c("ollama", "gemini", "none"),
  ...
)

Arguments

ct_obj	An htest object returned by stats::cor.test().
introduction	Optional character string giving the study context.
alpha	Numeric significance level (default 0.05).
audience	One of c("beginner","applied","advanced").
summary_only	Logical; if TRUE, return a 3-bullet executive summary regardless of audience depth (uses trainer_core_summary_only_block()).
llm_model	Character; model name for the generator (default "llama3").
generate	Logical; if TRUE, call the generator and return prompt + response.
llm_engine	Character; backend engine: "ollama", "gemini", or "none".
...	Passed to the selected LLM backend when 'generate = TRUE'.

Value

An 'entrainer_prompt' object. It behaves like a character string for 'cat()'/printing and stores LLM metadata and response as attributes when 'generate = TRUE'.

Privacy

If 'generate = TRUE' and 'llm_engine' is not '"none"', the prompt is sent to the selected LLM backend. With external providers such as Gemini, this may include excerpts of statistical outputs and user-provided context.

Examples

set.seed(1)
x <- rnorm(30); y <- 0.5*x + rnorm(30, sd = 0.8)
ct <- cor.test(x, y, method = "pearson")
cat(trainer_cor_test(ct, audience = "applied", summary_only = FALSE))

---

Deprecated short aliases for EntraineR trainers

Description

These aliases are kept for backward compatibility only. Prefer the explicit snake_case API: 'trainer_chisq_test()', 'trainer_cor_test()', 'trainer_prop_test()', and 'trainer_var_test()'.

Usage

trainer_chisq(...)

trainer_cor(...)

trainer_prop(...)

trainer_var(...)

Arguments

...	Passed to the corresponding trainer function.

Value

Same return value as the corresponding trainer.

---

Trainer: Interpret FactoMineR::LinearModel with an LLM-ready prompt

Description

Builds an English-only, audience-tailored prompt to interpret a FactoMineR::LinearModel result. Handles model selection (AIC/BIC) and instructs how to interpret deviation contrasts (sum-to-zero) for factors. Works for ANOVA, ANCOVA, and multiple regression.

Usage

trainer_linear_model(
  x,
  introduction = NULL,
  alpha = 0.05,
  t_test = NULL,
  audience = c("beginner", "applied", "advanced"),
  summary_only = FALSE,
  llm_model = "llama3",
  generate = FALSE,
  llm_engine = c("ollama", "gemini", "none"),
  ...
)

trainer_LinearModel(lm_obj, ...)

Arguments

x	An object returned by FactoMineR::LinearModel(...).
introduction	Optional character string giving the study context.
alpha	Numeric significance level (default 0.05).
t_test	Optional character vector to filter the T-test section by factor names and/or interactions (e.g. "FactorA" or "FactorA:FactorB").
audience	One of c("beginner","applied","advanced").
summary_only	Logical; if TRUE, return a 3-bullet executive summary.
llm_model	Character model name for the generator (e.g., "llama3").
generate	Logical; if TRUE, call the generator.
llm_engine	Character; backend engine: "ollama", "gemini", or "none".
...	Passed to the selected LLM backend when 'generate = TRUE'.
lm_obj	Deprecated name for 'x'.

Value

An 'entrainer_prompt' object. It behaves like a character string for 'cat()'/printing and stores LLM metadata and response as attributes when 'generate = TRUE'.

Privacy

If 'generate = TRUE' and 'llm_engine' is not '"none"', the prompt is sent to the selected LLM backend. With external providers such as Gemini, this may include excerpts of statistical outputs and user-provided context.

Examples

# --- Example 1: multiple regression with selection (ham) -------------------
data(ham)
if (requireNamespace("FactoMineR", quietly = TRUE)) {
  intro_ham <- "A sensory analysis institute wants to know if it's possible to predict
  the overall liking of a ham from its sensory description.
  A trained panel used the following attributes to describe 21 hams:
  Juiciness, Crispy, Tenderness, Pasty, Fibrous, Salty, Sweet, Meaty,
  Seasoned, Metallic, Ammoniated, Fatty, Braised, Lactic.
  Afterward, an Overall Liking score was assigned to each of the hams."
  # collapse whitespace safely without extra packages
  intro_ham <- gsub("\n", " ", intro_ham)
  intro_ham <- gsub("\\s+", " ", intro_ham)

  res <- FactoMineR::LinearModel(`Overall liking` ~ ., data = ham, selection = "bic")
  pr  <- trainer_linear_model(res, introduction = intro_ham, audience = "advanced",
                             generate = FALSE)
  cat(pr)
}

# --- Example 2: interaction with a categorical factor (deforestation) ------
data(deforestation)
if (requireNamespace("FactoMineR", quietly = TRUE)) {
  intro_flume <- "The study's goal is to determine how river deforestation affects
  the relationship between water and air temperature.
  The dataset contains maximum air and water temperatures measured over
  28 ten-day periods before deforestation and 28 periods after deforestation.
  The main objective is to understand if and how the link between air and
  water temperature changes after deforestation."
  intro_flume <- gsub("\n", " ", intro_flume)
  intro_flume <- gsub("\\s+", " ", intro_flume)

  res <- FactoMineR::LinearModel(Temp_water ~ Temp_air * Deforestation,
                                 data = deforestation, selection = "none")
  pr  <- trainer_linear_model(res, introduction = intro_flume, audience = "advanced",
                             generate = FALSE)
  cat(pr)
}

---

Trainer: Name an MCA dimension (FactoMineR::MCA) with an LLM-ready prompt

Description

Builds an English-only, audience-tailored prompt to name and justify a Multiple Correspondence Analysis (MCA) dimension from a FactoMineR::MCA object. The function never invents numbers: it passes verbatim excerpts from summary(mca_obj) and FactoMineR::dimdesc() filtered at a given significance threshold proba, and instructs how to read and name the axis.

Usage

trainer_mca(
  x,
  dimension = 1L,
  proba = 0.05,
  introduction = NULL,
  audience = c("beginner", "applied", "advanced"),
  summary_only = FALSE,
  llm_model = "llama3",
  generate = FALSE,
  llm_engine = c("ollama", "gemini", "none"),
  ...
)

trainer_MCA(mca_obj, ...)

Arguments

x	A MCA object returned by FactoMineR::MCA().
dimension	Integer scalar; the dimension (component) to name (default 1).
proba	Numeric in (0,1]; significance threshold used by FactoMineR::dimdesc() to characterize the dimension (default 0.05).
introduction	Optional character string giving the study context. Defaults to a generic description.
audience	One of c("beginner","applied","advanced").
summary_only	Logical; if TRUE, return a compact 3-bullet executive summary (uses trainer_core_summary_only_block()).
llm_model	Character; model name for your generator backend (default "llama3").
generate	Logical; if TRUE, calls trainer_core_generate_or_return() and returns a list with prompt, response, and model. If FALSE, returns the prompt string.
llm_engine	Character; backend engine: "ollama", "gemini", or "none".
...	Passed to the selected LLM backend when 'generate = TRUE'.
mca_obj	Deprecated name for 'x'.

Value

An 'entrainer_prompt' object. It behaves like a character string for 'cat()'/printing and stores LLM metadata and response as attributes when 'generate = TRUE'.

Privacy

If 'generate = TRUE' and 'llm_engine' is not '"none"', the prompt is sent to the selected LLM backend. With external providers such as Gemini, this may include excerpts of statistical outputs and user-provided context.

Examples

## Not run: 
# Example: tea (FactoMineR)
if (requireNamespace("FactoMineR", quietly = TRUE)) {
  data(tea, package = "FactoMineR")
  res_mca <- FactoMineR::MCA(tea, quanti.sup = 19, quali.sup = 20:36, graph = FALSE)

  intro <- "A survey on tea consumption practices and contexts was summarized by MCA."
  intro <- gsub("\n", " ", intro); intro <- gsub("\\s+", " ", intro)

  # Applied audience
  prompt <- trainer_mca(res_mca,
                        dimension = 1,
                        proba = 0.01,
                        introduction = intro,
                        audience = "applied",
                        generate = FALSE)
  cat(prompt)

  res <- gemini_generate(prompt, compile_to = "html")
}

## End(Not run)

---

Trainer: Name a PCA dimension (FactoMineR::PCA) with an LLM-ready prompt

Description

Builds an English-only, audience-tailored prompt to name and justify a principal component (dimension) from a FactoMineR::PCA object. The function never invents numbers: it passes verbatim excerpts from 'summary(pca_obj)' (Individuals/Variables) and 'FactoMineR::dimdesc()' filtered at a given significance threshold 'proba', and instructs how to read and name the axis.

Usage

trainer_pca(
  x,
  dimension = 1L,
  proba = 0.05,
  introduction = NULL,
  audience = c("beginner", "applied", "advanced"),
  summary_only = FALSE,
  llm_model = "llama3",
  generate = FALSE,
  llm_engine = c("ollama", "gemini", "none"),
  ...
)

trainer_PCA(pca_obj, ...)

Arguments

x	A PCA object returned by FactoMineR::PCA().
dimension	Integer scalar; the dimension (component) to name (default 1).
proba	Numeric in (0,1]; significance threshold used by FactoMineR::dimdesc() to characterize the dimension (default 0.05).
introduction	Optional character string giving the study context. Defaults to a generic description.
audience	One of c("beginner","applied","advanced").
summary_only	Logical; if TRUE, return a compact 3-bullet executive summary (uses trainer_core_summary_only_block()).
llm_model	Character; model name for your generator backend (default "llama3").
generate	Logical; if TRUE, calls trainer_core_generate_or_return() and returns a list with prompt, response, and model. If FALSE, returns the prompt string.
llm_engine	Character; backend engine: "ollama", "gemini", or "none".
...	Passed to the selected LLM backend when 'generate = TRUE'.
pca_obj	Deprecated name for 'x'.

Value

An 'entrainer_prompt' object. It behaves like a character string for 'cat()'/printing and stores LLM metadata and response as attributes when 'generate = TRUE'.

Privacy

If 'generate = TRUE' and 'llm_engine' is not '"none"', the prompt is sent to the selected LLM backend. With external providers such as Gemini, this may include excerpts of statistical outputs and user-provided context.

Examples

## Not run: 
# Example: decathlon (FactoMineR)
if (requireNamespace("FactoMineR", quietly = TRUE)) {
data(decathlon, package = "FactoMineR")

res_pca <- FactoMineR::PCA(decathlon,
                quanti.sup = 11:12,
                quali.sup = 13,
                graph = FALSE)

intro <- "A study was conducted on decathlon athletes.
Performances on each event were measured and summarized by PCA."
intro <- gsub("\n", " ", intro); intro <- gsub("\\s+", " ", intro)

prompt <- trainer_pca(res_pca,
                dimension = 1,
                proba = 0.05,
                introduction = intro,
                audience = "applied",
                generate = FALSE)

cat(prompt)

res <- gemini_generate(prompt, compile_to = "html")
}

## End(Not run)

---

Interpret a proportion test (prop.test) with an audience-aware LLM prompt

Description

Builds a clear, audience-tailored prompt to interpret stats::prop.test() results (one-sample vs target p, two-sample equality, k-group equality, or k-group vs given p). Aligned with other EntraineR trainers: no invented numbers; audience-specific guidance.

Usage

trainer_prop_test(
  pt_obj,
  introduction = NULL,
  alpha = 0.05,
  audience = c("beginner", "applied", "advanced"),
  summary_only = FALSE,
  llm_model = "llama3",
  generate = FALSE,
  llm_engine = c("ollama", "gemini", "none"),
  ...
)

Arguments

pt_obj	An htest object returned by stats::prop.test().
introduction	Optional character string giving the study context.
alpha	Numeric significance level (default 0.05).
audience	One of c("beginner","applied","advanced").
summary_only	Logical; if TRUE, return a 3-bullet executive summary regardless of audience depth (uses trainer_core_summary_only_block()).
llm_model	Character; model name for the generator (default "llama3").
generate	Logical; if TRUE, call the generator and return prompt + response.
llm_engine	Character; backend engine: "ollama", "gemini", or "none".
...	Passed to the selected LLM backend when 'generate = TRUE'.

Value

An 'entrainer_prompt' object. It behaves like a character string for 'cat()'/printing and stores LLM metadata and response as attributes when 'generate = TRUE'.

Privacy

If 'generate = TRUE' and 'llm_engine' is not '"none"', the prompt is sent to the selected LLM backend. With external providers such as Gemini, this may include excerpts of statistical outputs and user-provided context.

Examples

# One-sample
pt1 <- prop.test(x = 56, n = 100, p = 0.5)
cat(trainer_prop_test(pt1, audience = "beginner"))

# Two-sample
pt2 <- prop.test(x = c(42, 35), n = c(100, 90))
cat(trainer_prop_test(pt2, audience = "applied", summary_only = TRUE))

---

Interpret a Student's t-test (stats::t.test) with an LLM-ready prompt

Description

Builds a clear, audience-tailored prompt to interpret a base R stats::t.test() result. Identifies the test flavor (One-sample, Two-sample, Paired, Welch) and instructs the LLM to use ONLY printed values (p, t, df, CI, estimates) and avoid any new calculations.

Usage

trainer_t_test(
  tt_obj,
  introduction = NULL,
  alpha = 0.05,
  audience = c("beginner", "applied", "advanced"),
  summary_only = FALSE,
  llm_model = "llama3",
  generate = FALSE,
  llm_engine = c("ollama", "gemini", "none"),
  ...
)

Arguments

tt_obj	An htest object returned by stats::t.test().
introduction	Optional character string giving the study context in plain English.
alpha	Numeric significance level used for interpretation (default 0.05).
audience	One of c("beginner","applied","advanced").
summary_only	Logical; if TRUE, return a 3-bullet executive summary.
llm_model	Character; model name passed to your generator (default "llama3").
generate	Logical; if TRUE, call trainer_core_generate_or_return() and return prompt + response.
llm_engine	Character; backend engine: "ollama", "gemini", or "none".
...	Passed to the selected LLM backend when 'generate = TRUE'.

Value

An 'entrainer_prompt' object. It behaves like a character string for 'cat()'/printing and stores LLM metadata and response as attributes when 'generate = TRUE'.

Privacy

If 'generate = TRUE' and 'llm_engine' is not '"none"', the prompt is sent to the selected LLM backend. With external providers such as Gemini, this may include excerpts of statistical outputs and user-provided context.

Examples

set.seed(1)
tt1 <- t.test(rnorm(20, 0.1), mu = 0)              # one-sample
cat(trainer_t_test(tt1, audience = "beginner"))

x <- rnorm(18, 0); y <- rnorm(20, 0.3)
tt2 <- t.test(x, y, var.equal = FALSE)             # two-sample Welch
cat(trainer_t_test(tt2, audience = "applied", summary_only = TRUE))

---

Interpret an F test comparing two variances (var.test) with an audience-aware LLM prompt

Description

Builds a clear, audience-tailored prompt to interpret a base R stats::var.test() result.

Usage

trainer_var_test(
  vt_obj,
  introduction = NULL,
  alpha = 0.05,
  audience = c("beginner", "applied", "advanced"),
  summary_only = FALSE,
  llm_model = "llama3",
  generate = FALSE,
  llm_engine = c("ollama", "gemini", "none"),
  ...
)

Arguments

vt_obj	An htest object returned by stats::var.test().
introduction	Optional character string giving the study context.
alpha	Numeric significance level (default 0.05).
audience	One of c("beginner","applied","advanced").
summary_only	Logical; if TRUE, return a 3-bullet executive summary regardless of audience depth (uses trainer_core_summary_only_block()).
llm_model	Character; model name for the generator (default "llama3").
generate	Logical; if TRUE, call the generator and return prompt + response.
llm_engine	Character; backend engine: "ollama", "gemini", or "none".
...	Passed to the selected LLM backend when 'generate = TRUE'.

Value

An 'entrainer_prompt' object. It behaves like a character string for 'cat()'/printing and stores LLM metadata and response as attributes when 'generate = TRUE'.

Privacy

If 'generate = TRUE' and 'llm_engine' is not '"none"', the prompt is sent to the selected LLM backend. With external providers such as Gemini, this may include excerpts of statistical outputs and user-provided context.

Examples

set.seed(1)
x <- rnorm(25, sd = 1.0); y <- rnorm(30, sd = 1.3)
vt <- var.test(x, y)
cat(trainer_var_test(vt, audience = "applied"))
cat(trainer_var_test(vt, audience = "advanced", summary_only = TRUE))

---