[dipl] Perustieteiden korkeakoulu / SCI

Permanent URI for this collectionhttps://aaltodoc.aalto.fi/handle/123456789/21

Browse

Browsing [dipl] Perustieteiden korkeakoulu / SCI by Author "Aalto, Antti"

Filter results by typing the first few letters
Now showing 1 - 1 of 1
  • Predicting Complex Events in Sensor Data
    (2013) Kolehmainen, Jussi
     School of Science | Master's thesis
    Complex event processing (CEP) analyzes data streams and detects complicated situations in real-time. Domain experts write e ective EPL (event processing language) queries to de ne complex events that are detected. In combination with predictive analytics (PA), which uses mathematical models to predict the future, a framework for predicting complex events can be designed. In this thesis I describe how predictive event processing works and how a proof-of-concept framework can be built. As prediction tools I use two models: a distance-based model and a feature-based model. The former uses dynamic time warping (DTW) and k-nearest neighbour (kNN) algorithm while the latter employs wavelet analysis and support vector machines (SVMs). As an application of predictive complex event processing I consider house automation and present a real-life case study for the experimental section. The goal is to predict when a certain variable exceeds a limit value for a certain period of time. I also evaluate the performance of the system. With two variables, CO2 and VOC (volatile organic compounds), the first, distance-based model performs better with correct alarm rate of over 75 % and false alarm rate of under 10 %. The second, feature-based model turns out to be faster but more di cult to con gure properly. More meaningful complex events and more thorough time parameter optimization are suggested for future research.