SD-HOC: Seasonal Decomposition Algorithm for Mining Lagged Time Series

Publication Year: 2017 Publication Type : ConferenceProceeding


Mining time series data is a difficult process due to the lag factor and different time of data arrival. In this paper, we present Seasonal Decomposition for Human Occupancy Counting (SD-HOC), a customised feature transformation decomposition, novel way to estimate the number of people within a closed space using only a single carbon dioxide sensor. SD-HOC integrates time lag and line of best fit model in the preprocessing algorithms. SD-HOC utilises seasonal-trend decomposition with moving average to transform the preprocessed data and for each trend, seasonal and irregular component, different regression algorithms are modelled to predict each respective human occupancy component value. Utilising M5 method linear regression for trend and irregular component and dynamic time warping for seasonal component, a set of the prediction value for each component was obtained. Zero pattern adjustment model is infused to increase the accuracy and finally, additive decomposition is used to reconstruct the prediction value. The accuracy results are compared with other data mining algorithms such as decision tree, multi-layer perceptron, Gaussian processes - radial basis function, support vector machine, random forest, na¨ıve Bayes and support vector regression in two different locations that have different contexts.


    author = {Irvan Bastian Arief Ang and Flora D. Salim and Margaret Hamilton},
    bibsource = {dblp computer science bibliography,},
    biburl = {},
    booktitle = {Data Mining - 15th Australasian Conference, AusDM 2017, Melbourne, VIC, Australia, August 19-20, 2017, Revised Selected Papers},
    doi = {10.1007/978-981-13-0292-3_8},
    editor = {Yee Ling Boo and David Stirling and Lianhua Chi and Lin Liu and Kok-Leong Ong and Graham Williams},
    pages = {125--143},
    publisher = {Springer},
    series = {Communications in Computer and Information Science},
    timestamp = {Tue, 16 Jun 2020 01:00:00 +0200},
    title = {SD-HOC: Seasonal Decomposition Algorithm for Mining Lagged Time Series},
    url = {},
    volume = {845},
    year = {2017}


Related Publications

RUP: Large Room Utilisation Prediction with carbon dioxide sensor
Type : JournalArticle
Show More
A Scalable Room Occupancy Prediction with Transferable Time Series Decomposition of CO 2 Sensor Data
Type : JournalArticle
Show More
Topical Event Detection on Twitter
Type : ConferenceProceeding
Show More

© 2021 Flora Salim - CRUISE Research Group.