1: \begin{abstract}

2: The major contribution of this work is to propose a way to classify discrete paths of a solution of a stochastic differential equation, where the space-dependent drift coefficient characterizes the class.

3: We build a plug-in type classifier in the one dimensional framework, and prove its convergence assuming that a large learning sample of discrete paths observed on the same fixed time interval is available. A crucial point in our analysis is that the classifier is based on the nonparametric inference of the drift and diffusion coefficients. In particular, we are investigating the case where this estimation is performed on an interval which is growing to become the entire real line. 

4: \end{abstract}

5: