Stats.hpp

Go to the documentation of this file.

#pragma once
 
#include "utils/RangeVector.hpp"
#include "utils/Types.hpp"
 
#include <algorithm>
#include <cmath>
#include <map>
#include <set>
#include <stdexcept>
#include <vector>
#include <Eigen/Dense>
#include <pcg_random.hpp>

namespace ppforest2::stats {
  using RNG = pcg32;
 

  template<typename Y> void sort(types::FeatureMatrix& x, Y& y) {
    std::vector<int> indices = utils::range_vector(x.rows());
 
    std::stable_sort(indices.begin(), indices.end(), [&y](int i, int j) { return y(i) < y(j); });
 
    x = x(indices, Eigen::all).eval();
    y = y(indices, Eigen::all).eval();
  }

  std::set<types::GroupId> unique(types::GroupIdVector const& column);

  template<typename Derived> double var(Eigen::MatrixBase<Derived> const& data) {
    static_assert(
        Derived::ColsAtCompileTime == 1 || Derived::ColsAtCompileTime == Eigen::Dynamic,
        "var: expected a vector (single column)"
    );
 
    if (data.rows() == 0) {
      throw std::invalid_argument("var: data must have at least one row");
    }
 
    if (data.rows() == 1) {
      return 0.0;
    }
 
    double const mean = static_cast<double>(data.mean());
    return (data.array().template cast<double>() - mean).square().sum() / static_cast<double>(data.rows() - 1);
  }

  template<typename Derived> double sd(Eigen::MatrixBase<Derived> const& data) {
    return std::sqrt(var(data));
  }

  types::FeatureVector var(types::FeatureMatrix const& data);

  types::FeatureVector sd(types::FeatureMatrix const& data);

  types::Outcome majority_vote(std::vector<types::Outcome> const& preds);

  types::Outcome mean(std::vector<types::Outcome> const& preds);

  std::map<types::GroupId, int> group_indices(std::set<types::GroupId> const& groups);
}