KSL

KSL is a free (LGPL3) C++/Qt library with utilities for science. Our goal is to provide tools to facilitade and make fast the development of programs for data analisys and information retrieval. We start with simple array handling and linear regression as well as a fast and featureful chart widget for visualization.

KSL's (growing) list of current features is:

Data Visualization Widgets
Advanced Data Structures
Regression Engines

The numerical routines used Ksl classes are primarily from the Mighty GNU Scientific Library GSL We thank and congratulate it's developers for the great work. Ksl follows some of KDE's software design goals and all of it's philosophy but it is not an official KDE component.

PLOTTING

The following minimal code

#include <QApplication>
#include <Ksl/Chart.h>

using namespace Ksl;

int main(int argc, char *argv[]) {
    QApplication app(argc, argv);
    Chart chart;

    auto x = linspace(-2*M_PI, 2*M_PI, 0.2);
    chart.plot("Cos(X)", x, cos(x), "bl");

    chart.show();
    return app.exec();
}

produces a chart similar to the following

REGRESSION

Whith the following program predits building values at boston city from a set of known parameters by using a linear regression

#include <QApplication>
#include <QDebug>

#include <Ksl/Chart.h>
#include <Ksl/Csv.h>
#include <Ksl/MultiLineRegr.h>
using namespace Ksl;


int main(int argc, char *argv[])
{
	QApplication app(argc, argv);

	Chart chart;
	chart.scale()->axis(XYScale::LeftAxis)
		->setName("BUILDING VALUES");
	chart.scale()->axis(XYScale::BottomAxis)
		->setName("SAMPLE ORDER");


	// Open file containing data
	Csv csv("housing.data");
	if (csv.empty()) {
		cout << "CSV file not found";
		return -1;
	}

	// Perform regression
	MultiLineRegr regr(csv, linspace<int>(0, 12), 13);

	// create series of predicted values
	Array<1> y_data = csv.array(13);
	Array<1> y_model(y_data.size());

	for (int k=0; k<y_model.size(); k++)
		y_model[k] = regr.model(k);

	// plot comparison
	auto x = linspace(0.0, double(y_data.size()));
	chart.plot("value", x, y_data, "bl");
	chart.plot("fit value", x, y_model, "rl");

	chart.show();
	return app.exec();
}

The result is the following:

If you are looking for a C/Gtk solution for 2D visualization. Take a look at slope

elvismt/Ksl

KSL

PLOTTING

REGRESSION