Interpretable Time Series Classification using All-Subsequence Learning and Symbolic Representations in Time and Frequency Domains