CategoricalFeaturesBuilder< T > Struct Template Reference

#include <MLTableFunctions.hpp>

Public Member Functions
	CategoricalFeaturesBuilder (const ColumnList< TextEncodingDict > &cat_features, const ColumnList< T > &numeric_features, const int32_t cat_top_k, const float cat_min_fraction, const bool cat_include_others)
	CategoricalFeaturesBuilder (const ColumnList< TextEncodingDict > &cat_features, const int32_t cat_top_k, const float cat_min_fraction, const bool cat_include_others)
	CategoricalFeaturesBuilder (const ColumnList< TextEncodingDict > &cat_features, const ColumnList< T > &numeric_features, const std::vector< std::vector< std::string >> &cat_feature_keys)
	CategoricalFeaturesBuilder (const ColumnList< TextEncodingDict > &cat_features, const std::vector< std::vector< std::string >> &cat_feature_keys)
ColumnList< T >	getFeatures ()
const std::vector< std::vector < std::string > > &	getCatFeatureKeys () const

Private Attributes
int64_t	num_rows_
std::vector < TableFunctions_Namespace::OneHotEncoder_Namespace::OneHotEncodedCol < T > >	one_hot_encoded_cols_
std::vector< std::vector < std::string > >	cat_feature_keys_
std::vector< int8_t * >	col_ptrs_

Detailed Description

template<typename T>
struct CategoricalFeaturesBuilder< T >

Definition at line 385 of file MLTableFunctions.hpp.

Constructor & Destructor Documentation

template<typename T>

CategoricalFeaturesBuilder< T >::CategoricalFeaturesBuilder ( const ColumnList< TextEncodingDict > &  cat_features, const ColumnList< T > &  numeric_features, const int32_t  cat_top_k, const float  cat_min_fraction, const bool  cat_include_others  )

inline

Definition at line 387 of file MLTableFunctions.hpp.

References ColumnList< T >::numCols(), ColumnList< TextEncodingDict >::numCols(), and ColumnList< T >::ptrs_.

      : num_rows_(numeric_features.size()) {
    TableFunctions_Namespace::OneHotEncoder_Namespace::OneHotEncodingInfo
        one_hot_encoding_info(cat_top_k, cat_min_fraction, cat_include_others);
    const size_t num_cat_features = static_cast<size_t>(cat_features.numCols());
    std::vector<TableFunctions_Namespace::OneHotEncoder_Namespace::OneHotEncodingInfo>
        one_hot_encoding_infos;
    for (size_t cat_idx = 0; cat_idx < num_cat_features; ++cat_idx) {
      one_hot_encoding_infos.emplace_back(one_hot_encoding_info);
    }
    one_hot_encoded_cols_ =
        TableFunctions_Namespace::OneHotEncoder_Namespace::one_hot_encode<T>(
            cat_features, one_hot_encoding_infos);
    for (auto& one_hot_encoded_col : one_hot_encoded_cols_) {
      cat_feature_keys_.emplace_back(one_hot_encoded_col.cat_features);
      for (auto& one_hot_encoded_vec : one_hot_encoded_col.encoded_buffers) {
        col_ptrs_.emplace_back(reinterpret_cast<int8_t*>(one_hot_encoded_vec.data()));
      }
    }
    const int64_t num_numeric_features = numeric_features.numCols();
    for (int64_t numeric_feature_idx = 0; numeric_feature_idx < num_numeric_features;
         ++numeric_feature_idx) {
      col_ptrs_.emplace_back(numeric_features.ptrs_[numeric_feature_idx]);
    }
  }

Here is the call graph for this function:

template<typename T>

CategoricalFeaturesBuilder< T >::CategoricalFeaturesBuilder ( const ColumnList< TextEncodingDict > &  cat_features, const int32_t  cat_top_k, const float  cat_min_fraction, const bool  cat_include_others  )

inline

Definition at line 417 of file MLTableFunctions.hpp.

References ColumnList< TextEncodingDict >::numCols().

      : num_rows_(cat_features.size()) {
    TableFunctions_Namespace::OneHotEncoder_Namespace::OneHotEncodingInfo
        one_hot_encoding_info(cat_top_k, cat_min_fraction, cat_include_others);
    const size_t num_cat_features = static_cast<size_t>(cat_features.numCols());
    std::vector<TableFunctions_Namespace::OneHotEncoder_Namespace::OneHotEncodingInfo>
        one_hot_encoding_infos;
    for (size_t cat_idx = 0; cat_idx < num_cat_features; ++cat_idx) {
      one_hot_encoding_infos.emplace_back(one_hot_encoding_info);
    }
    one_hot_encoded_cols_ =
        TableFunctions_Namespace::OneHotEncoder_Namespace::one_hot_encode<T>(
            cat_features, one_hot_encoding_infos);
    for (auto& one_hot_encoded_col : one_hot_encoded_cols_) {
      cat_feature_keys_.emplace_back(one_hot_encoded_col.cat_features);
      for (auto& one_hot_encoded_vec : one_hot_encoded_col.encoded_buffers) {
        col_ptrs_.emplace_back(reinterpret_cast<int8_t*>(one_hot_encoded_vec.data()));
      }
    }
  }

Here is the call graph for this function:

template<typename T>

CategoricalFeaturesBuilder< T >::CategoricalFeaturesBuilder ( const ColumnList< TextEncodingDict > &  cat_features, const ColumnList< T > &  numeric_features, const std::vector< std::vector< std::string >> &  cat_feature_keys  )

inline

Definition at line 441 of file MLTableFunctions.hpp.

References ColumnList< T >::numCols(), ColumnList< TextEncodingDict >::numCols(), and ColumnList< T >::ptrs_.

      : num_rows_(numeric_features.size()), cat_feature_keys_(cat_feature_keys) {
    const size_t num_cat_features = static_cast<size_t>(cat_features.numCols());
    if (num_cat_features != cat_feature_keys_.size()) {
      throw std::runtime_error(
          "Number of provided categorical features does not match number of categorical "
          "features in the model.");
    }
    std::vector<TableFunctions_Namespace::OneHotEncoder_Namespace::OneHotEncodingInfo>
        one_hot_encoding_infos;
    for (size_t cat_idx = 0; cat_idx < num_cat_features; ++cat_idx) {
      one_hot_encoding_infos.emplace_back(cat_feature_keys_[cat_idx]);
    }
    one_hot_encoded_cols_ =
        TableFunctions_Namespace::OneHotEncoder_Namespace::one_hot_encode<T>(
            cat_features, one_hot_encoding_infos);
    for (auto& one_hot_encoded_col : one_hot_encoded_cols_) {
      for (auto& one_hot_encoded_vec : one_hot_encoded_col.encoded_buffers) {
        col_ptrs_.emplace_back(reinterpret_cast<int8_t*>(one_hot_encoded_vec.data()));
      }
    }
    const int64_t num_numeric_features = numeric_features.numCols();
    for (int64_t numeric_feature_idx = 0; numeric_feature_idx < num_numeric_features;
         ++numeric_feature_idx) {
      col_ptrs_.emplace_back(numeric_features.ptrs_[numeric_feature_idx]);
    }
  }

Here is the call graph for this function:

template<typename T>

CategoricalFeaturesBuilder< T >::CategoricalFeaturesBuilder ( const ColumnList< TextEncodingDict > &  cat_features, const std::vector< std::vector< std::string >> &  cat_feature_keys  )

inline

Definition at line 472 of file MLTableFunctions.hpp.

References ColumnList< TextEncodingDict >::numCols().

      : num_rows_(cat_features.size()), cat_feature_keys_(cat_feature_keys) {
    const size_t num_cat_features = static_cast<size_t>(cat_features.numCols());
    if (num_cat_features != cat_feature_keys_.size()) {
      throw std::runtime_error(
          "Number of provided categorical features does not match number of categorical "
          "features in the model.");
    }
    std::vector<TableFunctions_Namespace::OneHotEncoder_Namespace::OneHotEncodingInfo>
        one_hot_encoding_infos;
    for (size_t cat_idx = 0; cat_idx < num_cat_features; ++cat_idx) {
      one_hot_encoding_infos.emplace_back(cat_feature_keys_[cat_idx]);
    }
    one_hot_encoded_cols_ =
        TableFunctions_Namespace::OneHotEncoder_Namespace::one_hot_encode<T>(
            cat_features, one_hot_encoding_infos);
    for (auto& one_hot_encoded_col : one_hot_encoded_cols_) {
      for (auto& one_hot_encoded_vec : one_hot_encoded_col.encoded_buffers) {
        col_ptrs_.emplace_back(reinterpret_cast<int8_t*>(one_hot_encoded_vec.data()));
      }
    }
  }

Here is the call graph for this function:

Member Function Documentation

template<typename T>

const std::vector<std::vector<std::string> >& CategoricalFeaturesBuilder< T >::getCatFeatureKeys ( ) const

inline

Definition at line 502 of file MLTableFunctions.hpp.

Referenced by decision_tree_reg_fit__cpu_template(), gbt_reg_fit__cpu_template(), linear_reg_fit__cpu_template(), pca_fit__cpu_1(), pca_fit__cpu_template(), and random_forest_reg_fit__cpu_template().

                                                                   {
    return cat_feature_keys_;
  }

Here is the caller graph for this function:

template<typename T>

ColumnList<T> CategoricalFeaturesBuilder< T >::getFeatures ( )

inline

Definition at line 497 of file MLTableFunctions.hpp.

Referenced by decision_tree_reg_fit__cpu_template(), gbt_reg_fit__cpu_template(), linear_reg_fit__cpu_template(), pca_fit__cpu_1(), pca_fit__cpu_template(), and random_forest_reg_fit__cpu_template().

                              {
    return ColumnList<T>(
        col_ptrs_.data(), static_cast<int64_t>(col_ptrs_.size()), num_rows_);
  }

Here is the caller graph for this function:

Member Data Documentation

template<typename T>

std::vector<std::vector<std::string> > CategoricalFeaturesBuilder< T >::cat_feature_keys_

private

Definition at line 510 of file MLTableFunctions.hpp.

template<typename T>

std::vector<int8_t*> CategoricalFeaturesBuilder< T >::col_ptrs_

private

Definition at line 511 of file MLTableFunctions.hpp.

template<typename T>

int64_t CategoricalFeaturesBuilder< T >::num_rows_

private

Definition at line 507 of file MLTableFunctions.hpp.

template<typename T>

std::vector<TableFunctions_Namespace::OneHotEncoder_Namespace::OneHotEncodedCol<T> > CategoricalFeaturesBuilder< T >::one_hot_encoded_cols_

private

Definition at line 509 of file MLTableFunctions.hpp.

---

The documentation for this struct was generated from the following file:

• /home/jenkins-slave/workspace/core-os-doxygen/QueryEngine/TableFunctions/SystemFunctions/os/MLTableFunctions.hpp