Scilab provides tools to perform data analysis and modeling.
median — median (row median, column median,…) of vector/matrix/array entries
stdev — standard deviation (row or column-wise) of vector/matrix entries
variance — variance (and mean) of a vector or matrix (or hypermatrix) of real or complex numbers
Example with the binomial distribution probabilities:
More on wikipedia
Linear and nonlinear modeling
The simplest example is a Linear regression:
x = -30:30; y = x.^3; [a, b] = reglin(x, y); plot(x, y, "red") plot(x, a*x+b)
Scilab provides tools to perform predictive analytics and machine learning for both:
- supervised learning
- unsupervised learning
Supervised learning is a most common kind of machine learning approach. It basically works with a set of “right answers” that is being used for the learning phase.
Unsupervised learning on the opposite does not require to have a training set a labeled data. It automatically identifies the hidden structure behind the dataset, and group them in a certain number of classes defined by the user.
A regression problem deals with the prediction of a continuous function.
This example coming from the Scilab demos is a polynomial regression:
0001 function y=FF(x) 0002 // parametric function model 0003 y = a*(x-b)+c*x.*x; 0004 endfunction --> exec("SCI/modules/optimization/demos/datafit/datafit.dem.sce");
A classification problem predicts discrete-valued outputs of a function.