With the evolution of data and increasing popularity of IoT (Internet of Things), stream data mining has gained immense popularity. Researchers and developers are trying to analyze data patterns obtained from various devices. Stream data have several characteristics, the most important being its huge volume and high velocity. Although, a lot of research is being conducted in order to develop more efficient stream data mining techniques, pre-processing of stream data is an area that is under-studied. Real time applications generate data which is rather noisy and contain missing values. Apart from this, there is the issue of data evolution, which is a concern when dealing with stream data. To deal with the evolution of data, the proposed solution offers a hybrid of preprocessing techniques which are adaptive in nature. As a result of the study, an adaptive preprocessing and learning approach is implemented. The case study with sensor weather data demonstrates the results and accuracy of the proposed solution.
[1]
Bogdan Gabrys,et al.
Adaptive Preprocessing for Streaming Data
,
2014,
IEEE Transactions on Knowledge and Data Engineering.
[2]
Roshani Ade,et al.
Preprocessing of Streaming Data using Genetic Algorithm
,
2015
.
[3]
H. Ruda,et al.
Adaptive preprocessing for on-line learning with adaptive resonance theory (ART) networks
,
1995,
Proceedings of 1995 IEEE Workshop on Neural Networks for Signal Processing.
[4]
Bogdan Gabrys,et al.
Architecture for development of adaptive on-line prediction models
,
2009,
Memetic Comput..
[5]
Geoff Holmes,et al.
New ensemble methods for evolving data streams
,
2009,
KDD.
[6]
Piotr Duda,et al.
On Pre-processing Algorithms for Data Stream
,
2012,
ICAISC.
[7]
Saso Dzeroski,et al.
Learning model trees from evolving data streams
,
2010,
Data Mining and Knowledge Discovery.