Principal Component Analysis Visualization and State Discovery with Soil Data

Data exploration helps to gain understanding of the dataset and the system itself. There are methodologies to handle large number of sensors as well. In this paper operational states are defined to interpret physical behaviour in a soil ecosystem. Dimensionality reduction is achieved with Principal Component Analysis (PCA) method giving another view to the soil dataset from spring term. K-means algorithm groups data densities by clustering the data. This grouping is the basis for defining operational states in the system. Soil data as a part of an ecosystem involves specific features. In the applied approach dynamic visualization including animations constitute an important exploration view. All experiments are realized in Jupyter programming environment with Python 3 programming language. Related literature about data visualization is reviewed. Combining methods and tools with this data as a result soil ecosystem features are recognized.