The volume of time series data being gathered has significantly grown alongside the widespread use of sensing devices and the advancement of storage capabilities. This has led to the production of extensive time series datasets in recent years. To effectively analyze these datasets, considerable efforts are being dedicated to implementing temporal rule mining algorithms. By combining these approaches with the expertise of domain specialists, valuable insights can be derived from the interpretation of time series data, leading to significant findings about the underlying systems. Concurrently, numerous algorithms have been proposed to tackle the task of time series classification. Our objective is to enhance the interpretability of existing time series classification methods by bridging the gap between rule discovery and classification in time series mining. To achieve this, we introduce a novel algorithm called rule transform (RT). RT employs temporal algebra to convert a time series dataset into a new feature space composed of the support of the most salient temporal rules. This enables the generated feature space to be 1) qualitatively interpreted by domain experts and 2) utilized for classification purposes.  In this poster, we construct a multivariate time series dataset for Solar Energetic Particle (SEP) prediction using GOES proton and electron flux data and use RT to retrieve valuable insights.