Local structures, namely characteristic motifs, or prominent, nonliterally repeated patterns, play an important role in folk music. This paper uses Principal Component Analysis (PCA) to better understand characteristics of musical patterns and to further use this information for designing and evaluating future pattern discovery algorithms. We show what features can summarise the data variance in musical patterns and propose using feature selection and extraction methods to improve pattern discovery algorithms. Using PCA, we show the prominent features of MTC-ANN patterns. The pitch related and rhythmic features contribute together to the first PCA component; the second and third component consists mainly of pitch-related features and rhythmic features respectively. According to what PCA shows, in designing and evaluating pattern discovery algorithms, we should take metric structures into consideration as well as the repetitions and pitch related features in the patterns.