The smart home data analysis can be divided into two parts; one, domain is activity recognition that has been discussed in the last chapter, and the other one is wellness pattern generation and forecasting. The forecasting in the WSN based smart home is the dynamic learning from the historical sensing events. Transformation of prior sensing events into pattern and forecast can be done by the analysis of knowledge discovery and soft computing techniques. There are a number of knowledge and soft computing methods available, but these methods do not perform well in the AAL environment. Either these methods are complex and needs large training data or too simple where they offer poor accuracy (Moutacalli et al. 2015; Pulsford et al. 2011; Candas et al. 2014). For the Wellness Protocol based AAL the time series approach has been proposed and implemented. This time series approach includes the seasonal parameters from last year; it does not demand too much learning data. The rest of the chapter includes the wellness forecasting analysis and comparative results with other existing data mining methods.
[1]
S. Mukhopadhyay,et al.
Activity and Anomaly Detection in Smart Home: A Survey
,
2016
.
[2]
Dong-Soo Kwon,et al.
Unsupervised clustering for abnormality detection based on the tri-axial accelerometer
,
2009,
2009 ICCAS-SICE.
[3]
F. Kinnafick,et al.
Actigraph Accelerometer-Defined Boundaries for Sedentary Behaviour and Physical Activity Intensities in 7 Year Old Children
,
2011,
PloS one.
[4]
Abdenour Bouzouane,et al.
The behavioral profiling based on times series forecasting for smart homes assistance
,
2015,
J. Ambient Intell. Humaniz. Comput..
[5]
Víctor Peláez,et al.
An automatic data mining method to detect abnormal human behaviour using physical activity measurements
,
2014,
Pervasive Mob. Comput..
[6]
Jie Liu,et al.
Wellness Sensor Networks: A Proposal and Implementation for Smart Home for Assisted Living
,
2015,
IEEE Sensors Journal.