Deborah.DeborahCore.MLSequence

ml_sequence(
    cfg_model::String,
    ML_inputs::MLInputPreparer.MLInputBundle,
    partition::DatasetPartitioner.DatasetPartitionInfo,
    paths::PathConfigBuilderDeborah.DeborahPathConfig,
    X_list::Vector{String},
    jobid::Union{Nothing, String}=nothing
) -> Dict{Symbol, Matrix}

Main ML execution dispatcher for Deborah.DeborahCore. Chooses the appropriate model backend (e.g., LightGBM, Lasso, Ridge, etc.) based on cfg_model, and runs training $+$ prediction to generate output matrices for each configuration group.

Arguments

• cfg_model::String : Model name string (e.g., "LightGBM", "Lasso", "PyGBM", etc.)
• ML_inputs::MLInputPreparer.MLInputBundle : Preprocessed input data bundle ($X$, $Y$ vectors and indices).
• partition::DatasetPartitioner.DatasetPartitionInfo : Partitioning metadata (lb/tr/bc/ul counts and indices).
• paths::PathConfigBuilderDeborah.DeborahPathConfig : File path configuration (output directory, info file, etc.).
• X_list::Vector{String} : List of input feature keys.
• jobid::Union{Nothing, String} : Optional job ID used for logging.

Returns

• Dict{Symbol, Matrix} : A dictionary containing predicted $Y$ matrices for each group.
  • Keys: :YP_tr, :YP_bc, :YP_ul
  • Each value is a matrix of size $(N_\text{cfg}, N_\text{src})$ reconstructed from prediction vectors.