Time Series Data Mining Tool(TSDM Tool) A time series is a sequence of data points, measured typically at successive points in time spaced at uniform time intervals. Time series analysis comprises methods for analyzing time series data in order to extract meaningful statistics and other characteristics of the data.In the context of statistics,the primary goal of time series analysis is forecasting. In the context of signal processing it is used for signal detection and estimation, while in the context of data mining, pattern recognition and machine learning time series analysis can be used for clustering, classi cation, query by content, anomaly detection as well as forecasting. This project is aimed making a time series data mining tool which can be used to accomplish the above goals.
This project mainly focuses on analyzing the sea and rainfall level time series. The data sets considered belong to the rainfall data collected over ten years in the six taluks of Chikkaballapura district of Karnataka. The tool developed can be used to perform anomaly detection, forecasting, similarity detection and temporal pattern detection. The performance of these algorithms were tested on the above data sets and the results are presented. The tool is developed using the model view control design pattern. The algorithms are coded using Java using the object oriented paradigm. A web based interface with chart visualisation is provided for the end user. This is done using Java Server pages and Servlets. The aid of Google Charts API is taken for plotting graphs. The tool used Git revision control system and Github for online collaboration and code hosting.
The algorithms implemented in this tool require a set of user-de ned parameters that determine the accuracy of the results. The CUSUM and Statistical approach in the Anomaly-Detection module discover anomalies in the data sets. The Temporal Pattern Mining tool uses a tness threshold set by the user and shows temporal patterns, and similarly, the Dynamic Time Warping tool in the Similarity Module shows similarities among the time series data sets. The Neural Network in the Forecasting module is the most accurate among the algorithms with 60% accuracy.