Power quality information is useful not only to determine electricity fitness but also for another important application: load identification. One approach to identify loads is through non-intrusive load monitoring systems that estimate the nature and operation of individual appliances with measurements in a single point. In this work, load identification is addressed as a classification task by taking advantage of power quality information. Therefore, discriminant analysis of power quality characteristics are proposed, thus requiring simpler design and fitting processes than traditional techniques. In this regard, classifiers based on linear and quadratic discriminant analysis are implemented for laboratory measurements with noticeable performance.