README

This project is part of the course 'Advanced Topics in Machine Learning' at the OvGU Magdeburg.
Its objective is to investigate the suitability of different classifiers with regards to an activity recognition dataset:
http://archive.ics.uci.edu/ml/datasets/Heterogeneity+Activity+Recognition

Our approach utilizes Sliding Window: Since the data is recorded sequentially over time, this
information has to somehow be extracted and used by the classifier of choice. To achieve that
we group a given number of training instances together and use mean as well as standard deviation
with respect to the coordinates of the sensor. The class label for these new instances is decided
by majority vote.



Team members:
	Hardik Balar
	Noravee Sungpuag
	Florian Bethe

Supervisor:
	Marcus Thiel