Class for calculating statistics from CSV files and in-memory sources. More...

#include <csv_stat.hpp>

Public Types
using	FreqCount = std::unordered_map< std::string, size_t >

using	TypeCount = std::unordered_map< DataType, size_t >

Public Member Functions
std::vector< long double >	get_mean () const
	Return current means.

std::vector< long double >	get_variance () const
	Return current variances.

std::vector< long double >	get_mins () const
	Return current mins.

std::vector< long double >	get_maxes () const
	Return current maxes.

std::vector< FreqCount >	get_counts () const
	Get counts for each column.

std::vector< TypeCount >	get_dtypes () const
	Get data type counts for each column.

std::vector< std::string >	get_col_names () const

	CSVStat (csv::string_view filename, CSVFormat format=CSVFormat::guess_csv())
	Calculate statistics for an arbitrarily large file.

	CSVStat (std::stringstream &source, CSVFormat format=CSVFormat())
	Calculate statistics for a CSV stored in a std::stringstream.

Detailed Description

Class for calculating statistics from CSV files and in-memory sources.

Example

#include "csv.hpp"
 
int main(int argc, char** argv) {
    using namespace csv;
 
    if (argc < 2) {
        std::cout << "Usage: " << argv[0] << " [file]" << std::endl;
        exit(1);
    }
 
    std::string filename = argv[1];
    CSVStat stats(filename);
 
    auto col_names = stats.get_col_names();
    auto min = stats.get_mins(), max = stats.get_maxes(),
        means = stats.get_mean(), vars = stats.get_variance();
 
    for (size_t i = 0; i < col_names.size(); i++) {
        std::cout << col_names[i] << std::endl
            << "Min: " << min[i] << std::endl
            << "Max: " << max[i] << std::endl
            << "Mean: " << means[i] << std::endl
            << "Var: " << vars[i] << std::endl;
    }
 
    return 0;
}

Definition at line 18 of file csv_stat.hpp.

Member Typedef Documentation

◆ FreqCount

using csv::CSVStat::FreqCount = std::unordered_map<std::string, size_t>

Definition at line 20 of file csv_stat.hpp.

◆ TypeCount

using csv::CSVStat::TypeCount = std::unordered_map<DataType, size_t>

Definition at line 21 of file csv_stat.hpp.

Constructor & Destructor Documentation

◆ CSVStat() [1/2]

csv::CSVStat::CSVStat	(	csv::string_view	filename,
		CSVFormat	format = `CSVFormat::guess_csv()`
	)

Calculate statistics for an arbitrarily large file.

When this constructor is called, CSVStat will process the entire file iteratively. Once finished, methods like get_mean(), get_counts(), etc... can be used to retrieve statistics.

Definition at line 13 of file csv_stat.cpp.

◆ CSVStat() [2/2]

csv::CSVStat::CSVStat	(	std::stringstream &	source,
		CSVFormat	format = `CSVFormat()`
	)

Calculate statistics for a CSV stored in a std::stringstream.

Definition at line 19 of file csv_stat.cpp.

Member Function Documentation

◆ get_col_names()

std::vector< std::string > csv::CSVStat::get_col_names ( ) const

inline

Definition at line 30 of file csv_stat.hpp.

◆ get_counts()

std::vector< CSVStat::FreqCount > csv::CSVStat::get_counts ( ) const

Get counts for each column.

Definition at line 61 of file csv_stat.cpp.

◆ get_dtypes()

std::vector< CSVStat::TypeCount > csv::CSVStat::get_dtypes ( ) const

Get data type counts for each column.

Definition at line 70 of file csv_stat.cpp.

◆ get_maxes()

std::vector< long double > csv::CSVStat::get_maxes ( ) const

Return current maxes.

Definition at line 52 of file csv_stat.cpp.

◆ get_mean()

std::vector< long double > csv::CSVStat::get_mean ( ) const

Return current means.

Definition at line 25 of file csv_stat.cpp.

◆ get_mins()

std::vector< long double > csv::CSVStat::get_mins ( ) const

Return current mins.

Definition at line 43 of file csv_stat.cpp.

◆ get_variance()

std::vector< long double > csv::CSVStat::get_variance ( ) const

Return current variances.

Definition at line 34 of file csv_stat.cpp.

The documentation for this class was generated from the following files:

include/internal/csv_stat.hpp
include/internal/csv_stat.cpp

Public Types

Public Member Functions