The e-Home project from the Vienna University of Technology is an R&D project with goals of providing assistive technologies for private households of older people with the idea to give them possibilities for longer and independent living in their homes. The e-Home system consists of an adaptive intelligent network of wireless sensors for activity monitoring with a central context-aware embedded system.
The primary goal of this thesis is to investigate unsupervised prediction and clustering possibilities of user behaviour based on collected time-series data from infrared temperature sensors in the e-Home enviroment.
Three different prediction approaches are described. Hourly Based Event Binning approach is compared to two clustering algorithms, Hierarchical Clustering and Dirichlet Process GMM. Prediction rates are measured on data from three different test persons.
This thesis first examines two different approaches for event detection from infrared signal data. In a second stage three different methods for unsupervised prediction analytics are discussed and tested on selected data-sets. Clustering algorithms parameter settings for time-series data have also been discussed and tested in detail. Finally the prediction performance results are compared and each method's advantages and disadvantages have been discussed.
The practical part of this thesis is implemented in IPython notebook. Python version was 2.7 on 64 bit Ubuntu linux 12.04 LTS. Data analysis has been implemented with Python’s Pandas library. Visualisations are made with Matplotlib and Seaborn libraries.
The results reveal that prediction accuracy depends on data quantity and spread of data points. The simplest method in prediction comparison, the Hourly Based Binning has however given the best prediction rates overall.
The Dirichlet Process Gaussian Mixture Models clustering show best prediction performance on smaller training data sets and well spread data. By further parameter tuning on Dirichlet Process GMM clustering the prediction rates could be further improved coming very close or even over performing the Hourly Based Binning.
Due to the unknown distribution and well spread data, choosing the right threshold parameter for the Hierarchical Clustering was trickier than initially assumed. Despite the initial assumptions for Hierarchical Clustering, this method was at least applicable for unsupervised prediction analytics on used data sets.
Table of Contents
1. Introduction
1.1. Motivation
1.2. Outline
1.3. The Goal
1.4. Methodology
2. Theory Section
2.1. Gaussian Mixture Model
2.2. Dirichlet Process GMM
2.3. Hierarchical Clustering
3. Implementation
3.1. Software Specification
3.2. Data and Sensors
3.3. Data Visualisation and Inspection
3.4. Sensor values Discretisation and Extraction
3.5. Unsupervised Event Extraction
3.6. Data Structure for Event Analysis
3.7. Data Quality and Quantity
3.8. Predictive Analysis
3.8.1. Hourly Binning Analysis
3.8.2. Clustering Analysis
3.8.2.1. Hierarchical Clustering Analysis
3.8.2.2. Dirichlet Process Gaussian Mixture Model Clustering Analysis
4. Summary
4.1. Results
4.2. Discussion
Research Objectives and Themes
This thesis investigates unsupervised methods for predicting and clustering user behavior based on time-series data collected from infrared temperature sensors in an ambient assisted living (AAL) environment, with the aim of reliably identifying cooking activities.
- Adaptive intelligent sensor networks for elderly assistance.
- Unsupervised event detection from infrared signal data.
- Comparative analysis of prediction algorithms: Hourly Based Binning, Hierarchical Clustering, and Dirichlet Process GMM.
- Data preprocessing techniques including noise filtering, signal discretisation, and event sequencing.
- Evaluation of prediction performance based on data quality, quantity, and training set sizes.
Excerpt from the Book
3.4 Sensor Values Discretisation and Extraction
As noted earlier IIR sensor values need discretisation. The idea is to implement some kind of sampling on continuous values of the IIR cooking sensor. This is needed to be able to extract the cookings from the rest of the sensor events from the dataset.
Having a discrete signal from infrared temperature sensor, two approaches for cooking event extraction from a dataset are going to be discussed. The goal is to have possibilities of reliable and correct cooking events detection in a single dataset.
The first approach is the sequencing of temperature rises that belong together. The second approach is unsupervised event extraction using a clustering algorithm for grouping temperature increases that belong together. The latter is discussed in the next section.
Three big peaks from Figure 3.1 need to be discretized since only “heating in progress” on the cooking plate is of interest. This gives an idea that positive temperature increases should be inspected and leads to conclusion that the first step of temperature signal transforming should be a difference operation on temperature values.
Summary of Chapters
1. Introduction: Presents the motivation for AAL systems and defines the goal of detecting elderly cooking patterns without supervised intervention.
2. Theory Section: Discusses the theoretical foundations of Gaussian Mixture Models, Dirichlet Process GMM, and Hierarchical Clustering.
3. Implementation: Describes the practical setup, sensor data formats, signal discretization, and the execution of predictive analysis using different clustering and binning approaches.
4. Summary: Evaluates the performance of the implemented methods and discusses findings regarding data quality and future improvements for adaptive systems.
Keywords
AAL, Data Cluster Analysis, Predictive Analysis, Gaussian Mixture Model, DPGMM, Hierarchical Clustering, Hourly Binning, Time-Series Data, Infrared Sensors, User Behaviour, Unsupervised Learning, Activity Monitoring, Sensor Data Discretisation, Event Extraction, Smart Homes
Frequently Asked Questions
What is the primary focus of this research?
The research focuses on analyzing time-series sensor data from private households to identify and predict cooking activities of older people to support independent living.
What are the main thematic fields addressed?
The work combines ambient assisted living (AAL) technologies, machine learning for clustering, and data preprocessing for predictive modeling.
What is the central research question?
The thesis aims to determine how reliable and unsupervised prediction of daily user behavior can be achieved using various clustering and binning algorithms on raw infrared sensor signals.
Which scientific methods are employed?
The author evaluates Hourly Based Binning against more complex unsupervised clustering techniques, specifically Hierarchical Clustering and Dirichlet Process Gaussian Mixture Models (DPGMM).
What does the implementation part cover?
It details the software environment (Python/IPython), data cleansing of noisy CSV inputs, event isolation, sequence naming, and the comparison of different algorithms against training and testing datasets.
Which criteria are used to evaluate the methods?
The methods are evaluated based on prediction hit rates, computational efficiency, robustness regarding sparse/noisy data, and the ability to find a realistic number of clusters.
How does the DPGMM approach compare to the other methods?
DPGMM shows robustness with smaller, sparse datasets, though it often requires parameter tuning to outperform the simpler Hourly Binning approach.
Why did the author use Hierarchical Clustering?
It was chosen as an unsupervised grouping method that requires fewer parameters than other models and serves as a baseline to pre-estimate cluster counts for more complex models like DPGMM.
What impact does data quality have on the results?
The study concludes that prediction performance is directly correlated with data density and recording duration, with smaller or noisier datasets leading to lower accuracy and higher miss rates.
- Quote paper
- Dzenan Hamzic (Author), 2016, AAL Data Cluster Analysis. Theory and Implementation, Munich, GRIN Verlag, https://www.grin.com/document/340200
