Assisting the End-User in the Interpretation of Profiles for Decision Support. An Application to Wastewater Treatment Plants

Abstract

This paper describes the integral Knowledge Discovery (KDD) process, including both prior expert knowledge and interpretation oriented tools to extract the behavior of a real pilot wastewater treatment plant. Special emphasis is made on the interest of developing postprocessing tools for clustering methods which can help the expert to understand the meaning of the clusters and bridge the important existing gap between Data Mining and effective Decision Support. Traffic Lights Panel (TLP) is presented as a suitable visual interpretation oriented tool for clustering results. Based on this tool, four typical behaviours are identified in the pilot plant, which have been validated by the experts. Till now, the TLP is manually derived from the clustering results, but it has been well accepted by the domain experts of several real applications as a very helpful contribution to understand the classes meaning and improve reliable decision-making. Here, a proposal for automatic construction of TLP is presented trying to mimic the real process that the analyst performs to manually build them. A criterion based on conditional Median as a central trend statistics of the variables inside a class is introduced and refined to gain robustness towards outliers. Both criteria are tested and compared with the real target case study. A deep analysis of the advantages and drawbacks of the proposed criterion, permitted to better understand the analyst process when manually building TLPs, to identify the scope of the proposal, and to typify some of the situations in which additional conditions are required.

Identifying Health Status of Wind Turbines by Using Self Organizing Maps and Interpretation-Oriented Post-Processing Tools

“…Columns correspond to variables and rows to clusters. Histograms or box-plots are displayed for numerical variables and bar-charts for qualitative ones [50]. It allows to identify particularities of classes in regards of specific variables.…”

“…Both of them were designed with the aim of helping experts to conceptualize and label the resulting classes. Originally, CPGs and TLPs [50] were designed in the context of hierarchical clustering. In this paper, for the first time, they are used on clusters induced from a SOM network.…”

“…One step forward in the level of abstraction of the interpretation-support tool is the TLP. TLP is a symbolic post-processing of the clustering results proved extremely useful and well-accepted by domain experts in several real applications [11,50]. TLP exploits the association between the traffic light colors and the main central trend of the variables in every class to help the expert to understand the clusters and to support the conceptualization.…”

aTLP: A color-based model of uncertainty to evaluate the risk of decisions based on prototypes

“…Traffic Lights Panel (TLP) is a symbolic postprocessing of the clustering results [11] proved extremely useful and well-accepted by domain experts in several real applications [8][9][10]. TLP exploits the association between the traffic light colours and the general trend of the variables in every class to help the expert to understand the clusters and to support the conceptualization of the discovered classes.…”

Positioning of sensors for control of ventilation systems in broiler houses: a case study

“…Green color codifies thermal comfort, yellow is for warning situation, and red color codifies dangerous conditions related to the THI index. This TLP association has been used as a potential tool to support decisions after statistical analysis and data mining applications (Gibert et al, 2008;Gibert et al, 2012;Gibert and Conti, 2014).…”