Providing a robust traffic management system is a given for road authorities today, for safer mobility and reliable day-to-day operations. Therefore, to improve traffic management procedures in order to provide an open road for road users at all times, understanding road surface friction measurement is crucial. Currently, the automatic prediction model of road surface friction measurement (FriC-PM) is highly demanded by road transport owners and operators. The prediction accuracy of road surface friction is linked to several external variables related to the weather conditions, road surface micro- and macro-texture, tire properties, driving speed and behaviour of road users. This paper presents a trained novel predictive model developed for the measurement of road surface friction considering a big dataset of 18 months with daily records through novel intelligent road-based passive sensor measurement, on a Spanish highway section. The trained predictive model is developed on the machine learning (ML) approaches, namely support vector machine (SVM), and validated with the K-Fold cross-validation (CV) algorithm taking into account various kernels. In addition, error evaluation strategies are applied to the trained model in order to identify the efficiency, accuracy, and reliability of the predictive model. The main purpose of the proposed prediction model is to provide comprehensive knowledge about predictive road surface friction coefficient, and to support decision-making elements to road owners or operators for reliable predictive maintenance process.