An Alternative Method For Random-Pattern Reading and Prediction of Nominal and Ordinal Time Series Data Type: Chain of Event Method

Abstract

To read the patterns and to predict the ordinal and nominal data type, a researcher can use several statistical method such as Markov Chain Analysis. Each of those statistical methods have a different approach. Chain of event method is the statistical method made by the author which can be used by researchers as an alternative method to predict and interpret the pattern of time series data type. This method uses a different approach if we compare it with the other methods, namely by reading a random pattern of historical data and analyze a series of events that are most likely to occur in the future. It aims to overcome the weaknesses of the methods that existed before. Because this method has two functions, namely for the prediction and reading patterns, this method is tested using a separate statistical tests. To test its pattern reading ability, the author is used the sign test with significance level of 0.05, one-tail, and the samples were 30 data. To test the accuracy of predictions, the author is using two tests, the first is Wilcoxon sum rank test with significance level of 0.01, one-tail, and with sample number of 30 data. The second test is sign test with significance level of 0.05, one-tail, and with the amount of 30 samples of data. The results of these tests is satisfactorily. From the results of these tests, can be concluded that the pattern reading ability and accuracy of chain of event method is satisfying.