module Numo
  module Libsvm
    module SvmType
      C_SVC: Integer
      NU_SVC: Integer
      ONE_CLASS: Integer
      EPSILON_SVR: Integer
      NU_SVR: Integer
    end

    module KenelType
      LINEAR: Integer
      POLY: Integer
      RBF: Integer
      SIGMOID: Integer
      PRECOMPUTED: Integer
    end

    LIBSVM_VERSION: Integer
    VERSION: String

    type model = {
      nr_class: Integer,
      l: Integer,
      SV: Numo::DFloat,
      sv_coef: Numo::DFloat,
      rho: Numo::DFloat,
      probA: Numo::DFloat,
      probB: Numo::DFloat,
      prob_density_marks: Numo::DFloat,
      sv_indices: Numo::Int32,
      label: Numo::Int32,
      nSV: Numo::Int32,
      free_sv: Integer
    }

    type param = {
      svm_type: Integer?,
      kernel_type: Integer?,
      degree: Integer?,
      gamma: Float?,
      coef0: Float?,
      cache_size: Float?,
      eps: Float?,
      C: Float?,
      nr_weight: Integer?,
      weight_label: Numo::Int32?,
      weight: Numo::DFloat?,
      nu: Float?,
      p: Float?,
      shrinking: bool?,
      probability: bool?,
      verbose: bool?,
      random_seed: Integer?
    }

    def self?.cv: (Numo::DFloat x, Numo::DFloat y, param, Integer n_folds) -> Numo::DFloat
    def self?.train: (Numo::DFloat x, Numo::DFloat y, param) -> model
    def self?.predict: (Numo::DFloat x, param, model) -> Numo::DFloat
    def self?.predict_proba: (Numo::DFloat x, param, model) -> Numo::DFloat
    def self?.decision_function: (Numo::DFloat x, param, model) -> Numo::DFloat
    def self?.save_svm_model: (String filename, param, model) -> bool
    def self?.load_svm_model: (String filename) -> [param, model]
  end
end