study.yaml schema

The declarative study config consumed by tamga run. A minimal example:

name: my-study
seed: 42

corpus:
  path: corpus
  metadata: corpus/metadata.tsv

features:
  - id: mfw200
    type: mfw
    n: 200
    scale: zscore
    lowercase: true

methods:
  - id: burrows
    kind: delta
    method: burrows
    features: mfw200
    group_by: author

Top-level keys

Key	Type	Required	Description
name	str	yes	Study name; shows in reports
seed	int	no	Default seed (42). Threaded to every stochastic method.
corpus	object	yes	Corpus config (below)
features	list	yes	One or more feature extractors
methods	list	yes	One or more methods to run
output	object	no	Output directory / timestamping
cache	object	no	DocBin cache directory
preprocess	object	no	spaCy model selection

corpus

corpus:
  path: corpus                    # directory of .txt files
  metadata: corpus/metadata.tsv   # optional TSV with filename + arbitrary fields
  strict: true                    # default: raise if any file lacks metadata
  filter:                         # optional: subset the corpus before running
    role: [train]

features

Each feature extractor is a dict with an id (referenced by methods), a type, and type-specific params.

Supported types

type	params
mfw	n, min_df, max_df, scale ({none, zscore, l1, l2}), lowercase
word_ngram	n (int or [min, max]), lowercase, scale
char_ngram	n, include_boundaries, scale
function_word	wordlist (optional list or path), scale
punctuation	(none)
lexical_diversity	(none)
readability	(none)

methods

Each method is a dict with an id, a kind, an optional features (feature id), plus params.

Supported kinds

kind	Description
delta	Nearest-centroid attribution (method: burrows by default)
zeta	Craig's Zeta; requires group_by and either inferred or specified params.group_a / group_b
reduce	Dim-reduction (default PCA); params.n_components
cluster	Hierarchical (default Ward); params.n_clusters, params.linkage
consensus	Bootstrap consensus tree; params.mfw_bands, params.replicates
classify	sklearn classifier; params.estimator, cv.kind, cv.folds

output

output:
  dir: results          # default
  timestamp: true       # wrap runs in timestamped subdirectories

cache

cache:
  dir: .tamga/cache     # spaCy DocBin cache location

preprocess

preprocess:
  spacy:
    model: en_core_web_trf    # default; change to sm/md for speed

A realistic multi-method example

name: federalist
seed: 42
output: { dir: results, timestamp: false }

corpus:
  path: corpus
  metadata: corpus/metadata.tsv
  filter:
    role: [train]

features:
  - id: mfw200
    type: mfw
    n: 200
    scale: zscore
    lowercase: true

methods:
  - id: burrows
    kind: delta
    method: burrows
    features: mfw200
    group_by: author

  - id: pca
    kind: reduce
    features: mfw200
    params: { n_components: 2 }

  - id: ward
    kind: cluster
    features: mfw200
    params: { n_clusters: 3, linkage: ward }

  - id: zeta_h_m
    kind: zeta
    group_by: author
    params:
      top_k: 50
      group_a: Hamilton
      group_b: Madison