Fast Classification of Time Series Data

Nowadays, we have to deal with fast-growing and permanently evolving data, including social networks and sensor data recorded from smart phones or vehicles. Data evolving in time brings a lot of new challenges to the data mining and machine learning community. In recent years, classification of time series data has become a topic of great interest within the data mining community. Several approaches or classifiers have been proposed for the problem of time series classification. We propose an approach which is computationally fast and accurate when compared with 1NN(One Nearest Neighbor) and kNN(k Nearest Neighbor) classifiers. Our approach speeds up the computation by restraining the search for the closest pattern only to a subset of classes. During classification phase we are using a window parameter in order to classify the test pattern using only a subset of the training patterns. We have tested our approach with a number of datasets and compared our approach with 1NN and kNN classifiers.