Outlier detection in weight time series of connected scales.

Saeed Mehrang; Elina Helander; Misha Pavel; Angela Chieh; Ilkka Korhonen

doi:10.1109/BIBM.2015.7359896

Outlier detection in weight time series of connected scales.

Saeed Mehrang, Elina Helander, Misha Pavel, Angela Chieh, Ilkka Korhonen

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review

26 Citations (Scopus)

Abstract

In principle, connected sensors allow effortless long-term self-monitoring of health and wellness that can help maintain health and quality of life. However, data collected in the ”wild” may be noisy and contain outliers, e.g., due to uncontrolled sources or data from different persons using the same device. The removal of the ”outliers” is therefore critical for accurate interpretation of the data. In this paper we study the detection and elimination of outliers in selfweighing time series data obtained from connected weight scales. We examined three techniques: (1) a method based on autoregressive integrated moving average (ARIMA) time series modelling, (2) median absolute deviation (MAD) scale estimate, and (3) a method based on Rosner statistics. We applied these methods to both a data set with real outliers and a clean data set corrupted with simulated outliers. The results suggest that the simple MAD algorithm and ARIMA performed well with both test sets while the Rosner statistics was significantly less effective. In addition, the ARIMA approach appeared to be significantly less sensitive to long periods of missing data than
MAD and Rosner statistics.

Original language	English
Title of host publication	2015 IEEE International Conference on Bioinformatics and Biomedicine (BIBM)
Subtitle of host publication	Second International Workshop on the Role for Quantified Self for Personal Healthcare (QSPH’15)
Publisher	IEEE
Pages	1489-1496
Number of pages	8
ISBN (Print)	978-1-4673-6799-8
DOIs	https://doi.org/10.1109/BIBM.2015.7359896
Publication status	Published - 2015
Publication type	A4 Article in conference proceedings
Event	IEEE International Conference on Bioinformatics and Biomedicine - , United States Duration: 1 Jan 2015 → …

Conference

Conference	IEEE International Conference on Bioinformatics and Biomedicine
Country/Territory	United States
Period	1/01/15 → …

Publication forum classification

No publication forum level

Access to Document

10.1109/BIBM.2015.7359896

Cite this

Mehrang, S., Helander, E., Pavel, M., Chieh, A., & Korhonen, I. (2015). Outlier detection in weight time series of connected scales. In 2015 IEEE International Conference on Bioinformatics and Biomedicine (BIBM): Second International Workshop on the Role for Quantified Self for Personal Healthcare (QSPH’15) (pp. 1489-1496). IEEE. https://doi.org/10.1109/BIBM.2015.7359896

@inproceedings{ddc08238f14c48b596dd7c9544120e32,

title = "Outlier detection in weight time series of connected scales.",

abstract = "In principle, connected sensors allow effortless long-term self-monitoring of health and wellness that can help maintain health and quality of life. However, data collected in the ”wild” may be noisy and contain outliers, e.g., due to uncontrolled sources or data from different persons using the same device. The removal of the ”outliers” is therefore critical for accurate interpretation of the data. In this paper we study the detection and elimination of outliers in selfweighing time series data obtained from connected weight scales. We examined three techniques: (1) a method based on autoregressive integrated moving average (ARIMA) time series modelling, (2) median absolute deviation (MAD) scale estimate, and (3) a method based on Rosner statistics. We applied these methods to both a data set with real outliers and a clean data set corrupted with simulated outliers. The results suggest that the simple MAD algorithm and ARIMA performed well with both test sets while the Rosner statistics was significantly less effective. In addition, the ARIMA approach appeared to be significantly less sensitive to long periods of missing data thanMAD and Rosner statistics.",

author = "Saeed Mehrang and Elina Helander and Misha Pavel and Angela Chieh and Ilkka Korhonen",

note = "AUX=sgn,{"}Mehrang, Saeed{"} ; IEEE International Conference on Bioinformatics and Biomedicine ; Conference date: 01-01-2015",

year = "2015",

doi = "10.1109/BIBM.2015.7359896",

language = "English",

isbn = "978-1-4673-6799-8",

pages = "1489--1496",

booktitle = "2015 IEEE International Conference on Bioinformatics and Biomedicine (BIBM)",

publisher = "IEEE",

address = "United States",

}

Mehrang, S, Helander, E, Pavel, M, Chieh, A & Korhonen, I 2015, Outlier detection in weight time series of connected scales. in 2015 IEEE International Conference on Bioinformatics and Biomedicine (BIBM): Second International Workshop on the Role for Quantified Self for Personal Healthcare (QSPH’15). IEEE, pp. 1489-1496, IEEE International Conference on Bioinformatics and Biomedicine, United States, 1/01/15. https://doi.org/10.1109/BIBM.2015.7359896

Outlier detection in weight time series of connected scales. / Mehrang, Saeed; Helander, Elina; Pavel, Misha et al.
2015 IEEE International Conference on Bioinformatics and Biomedicine (BIBM): Second International Workshop on the Role for Quantified Self for Personal Healthcare (QSPH’15). IEEE, 2015. p. 1489-1496.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review

TY - GEN

T1 - Outlier detection in weight time series of connected scales.

AU - Mehrang, Saeed

AU - Helander, Elina

AU - Pavel, Misha

AU - Chieh, Angela

AU - Korhonen, Ilkka

N1 - AUX=sgn,"Mehrang, Saeed"

PY - 2015

Y1 - 2015

N2 - In principle, connected sensors allow effortless long-term self-monitoring of health and wellness that can help maintain health and quality of life. However, data collected in the ”wild” may be noisy and contain outliers, e.g., due to uncontrolled sources or data from different persons using the same device. The removal of the ”outliers” is therefore critical for accurate interpretation of the data. In this paper we study the detection and elimination of outliers in selfweighing time series data obtained from connected weight scales. We examined three techniques: (1) a method based on autoregressive integrated moving average (ARIMA) time series modelling, (2) median absolute deviation (MAD) scale estimate, and (3) a method based on Rosner statistics. We applied these methods to both a data set with real outliers and a clean data set corrupted with simulated outliers. The results suggest that the simple MAD algorithm and ARIMA performed well with both test sets while the Rosner statistics was significantly less effective. In addition, the ARIMA approach appeared to be significantly less sensitive to long periods of missing data thanMAD and Rosner statistics.

AB - In principle, connected sensors allow effortless long-term self-monitoring of health and wellness that can help maintain health and quality of life. However, data collected in the ”wild” may be noisy and contain outliers, e.g., due to uncontrolled sources or data from different persons using the same device. The removal of the ”outliers” is therefore critical for accurate interpretation of the data. In this paper we study the detection and elimination of outliers in selfweighing time series data obtained from connected weight scales. We examined three techniques: (1) a method based on autoregressive integrated moving average (ARIMA) time series modelling, (2) median absolute deviation (MAD) scale estimate, and (3) a method based on Rosner statistics. We applied these methods to both a data set with real outliers and a clean data set corrupted with simulated outliers. The results suggest that the simple MAD algorithm and ARIMA performed well with both test sets while the Rosner statistics was significantly less effective. In addition, the ARIMA approach appeared to be significantly less sensitive to long periods of missing data thanMAD and Rosner statistics.

U2 - 10.1109/BIBM.2015.7359896

DO - 10.1109/BIBM.2015.7359896

M3 - Conference contribution

SN - 978-1-4673-6799-8

SP - 1489

EP - 1496

BT - 2015 IEEE International Conference on Bioinformatics and Biomedicine (BIBM)

PB - IEEE

T2 - IEEE International Conference on Bioinformatics and Biomedicine

Y2 - 1 January 2015

ER -

Outlier detection in weight time series of connected scales.

Abstract

Conference

Publication forum classification

Access to Document

Fingerprint

Cite this