Modelling Similarity for Comparing Physical Activity Profiles - A Data-Driven Approach

Abstract

Objective measurements of physical behaviour are an interesting research field from the public health and computer science perspective. While for public health research, measurements with a high quality and feasible setup is important, the analysis of and reasoning about the data is what we will present in this work. Our focus in this work is the comprehensive representation of physical behaviour throughout consecutive days and allowing to find subgroups in the population with similar physical activity levels. We have a unique data set of 4628 participants wearing tri-axial accelerometers for six days and will present a case-based reasoning (CBR) system that can find and compare similar activity profiles. In this work, we focus on creating a CBR model using myCBR and do initial experiments with the resulting system. We will introduce a data-driven approach for modelling local similarity measures. Eventually, in the experiments we will show that for the given data set, the CBR system outperforms a k-Nearest Neighbor regressor in finding most similar participants.

A Data-Driven Approach for Determining Weights in Global Similarity Functions

“…For the process of evaluation, we create a case base for each dataset where all the features are included. The local similarity measures are modeled using the interquartile ranges for a numerical feature (see [27] for details), and pair-wise similarity for a categorical feature. As a baseline system each dataset has been provided to a CBR engineer to model the global and local similarities manually.…”

“…The relevant attributes can often be determined in collaboration with experts or using data driven approaches, while the definition of initial similarity measures are more challenging. This task has been addressed by researchers before, and learning or deriving similarity measures is an active field in CBR research [11,27].…”

Similarity Measure Development for Case-Based Reasoning–A Data-Driven Approach

“…Our approach can be applied on any dataset to model the similarity measures. A more detailed evaluation of our approach can be found in [7] where we statistically evaluated its effectiveness using a public health domain dataset and showed that the CBR Fig. 4.…”

“…From there we look into the Q 1 and Q 3 , which indicate the majority spread of the attributes in the data set. In line with [1,7], we decided to take these values as reference points for determining the decrease in similarity.…”

Using Automated Feature Selection for Building Case-Based Reasoning Systems: An Example from Patient-Reported Outcome Measurements

“…These datasets were used to build CBR systems in a datadriven manner and as training data in the other two regression algorithms. In all the CBR models built for various target outcomes in this work, local similarity modelling of the attributes has been done in the same data-driven manner as presented in our previous work [25,26]. The individual features are weighted equally in the global similarity function.…”