Open AccessTime Series Optimization on Data Mining
Author(s) -
Relita Buaton,
Herman Mawengkang,
Muhammad Zarlis,
Syahril Effendi,
Akim Manaor Hara Pardede,
Yani Maulita,
Achmad Fauzi,
N Novriyenni
Publication year - 2019
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1235/1/012014
Subject(s) - series (stratigraphy) , computer science , data mining , time series , field (mathematics) , process (computing) , representation (politics) , abstraction , machine learning , mathematics , paleontology , philosophy , epistemology , politics , political science , pure mathematics , law , biology , operating system
Forecasting is one of the important topics in the data mining field, such as, predictions, weather forecasting, predictions of academic achievement. Another topic associated with forecasting through a series of data that depends on the time period is called time series. The problem in data mining time series is how to present the data. A common approach is to transform periodic series into other domains so that the reduced dimensions are followed by an index mechanism. However, research on time series has not been successful optimal yet, because it is still limited to mining data yet to represent time series, this pattern needs to be developed to change the pattern into a rule. The main problem that needs to be addressed in a periodic series is to present the results of visualization which includes more than thousands of observations are very difficult in order to present time series data in multidimensional to be mined. Working with high-dimensional data will be very expensive in terms of process and storage costs, because it requires high-level data representation or abstraction. The method used is the cluster window. The results obtained are the discovery of patterns based on the frequency of symbol behavior