Linear separability

What algorithm should be used to find the plane for a linearly separable dataset? I know that this is a quadratic programming problem, but I can’t find a suitable algorithm, could you please help?

Problem:

X = $$\{x_1, ..., x_n\}$$ - dataset, $$Y = \{y_1, ..., y_n\}$$ - classes. $$y_i \in \{1, -1\}, x_i \in \mathbb{R^n}$$. You should find vector $$(a_1, ..., a_n, b)$$ : $$\forall i = \overline{1,n} \ sign((a_i, x_i) + b) = y_i$$.

• you tagged it with "svm," though! – Taylor Jul 14 '19 at 20:12
• I might be missing something! But why wouldn't you use the logistic regression? Its definition is to separate linearly separable data points. – aghd Jul 14 '19 at 20:36
• The two most popular methods would be logistic regression (followed by appropriate thresholding of the predicted probabilities) and support vector machines. – user20160 Jul 14 '19 at 21:53

1 Answer

I would go for linear discriminate analysis , a method used in statistics, pattern recognition and machine learning to find a linear combination of features that characterizes or separates two or more classes of objects or events. The resulting combination may be used as a linear classifier.