Capturing waste collection planning expert knowledge in a fitness function through preference learning

TL;DR

This paper tackles unifying waste collection planning by learning a fitness function from expert preferences rather than relying on handcrafted rules. It designs 21 additive KPIs to describe routes (and their three stretches), generates alternative routes with a greedy method, and learns a linear ranking model f(z)=<w,z> from expert judgments. The approach demonstrates that feature selection reveals strong redundancy among KPIs, with truck load E and route-distance components driving the most predictive power, achieving a C-index of up to ~98% and outperforming single-KPI and PSO-based baselines. The work provides a practical pathway to integrate preference-based fitness evaluation into global optimization for municipal waste collection, with implications for improved resource usage and planning efficiency.

Abstract

This paper copes with the COGERSA waste collection process. Up to now, experts have been manually designed the process using a trial and error mechanism. This process is not globally optimized, since it has been progressively and locally built as council demands appear. Planning optimization algorithms usually solve it, but they need a fitness function to evaluate a route planning quality. The drawback is that even experts are not able to propose one in a straightforward way due to the complexity of the process. Hence, the goal of this paper is to build a fitness function though a preference framework, taking advantage of the available expert knowledge and expertise. Several key performance indicators together with preference judgments are carefully established according to the experts for learning a promising fitness function. Particularly, the additivity property of them makes the task be much more affordable, since it allows to work with routes rather than with route plannings. Besides, a feature selection analysis is performed over such indicators, since the experts suspect of a potential existing (but unknown) redundancy among them. The experiment results confirm this hypothesis, since the best $C-$index ($98\%$ against around $94\%$) is reached when 6 or 8 out of 21 indicators are taken. Particularly, truck load seems to be a highly promising key performance indicator, together to the travelled distance along non-main roads. A comparison with other existing approaches shows that the proposed method clearly outperforms them, since the $C-$index goes from $72\%$ or $90\%$ to $98\%$.

Figures (4)

• Figure 1: $C-$index for expert feature selection, parametric RFE and automatic RFE when SVM is taken as an algorithm that produces linear models
• Figure 2: $C-$index for expert feature selection, parametric RFE and automatic RFE when LR is taken as an algorithm that produces linear models
• Figure 3: $C-$index for expert feature selection, parametric RFE and automatic RFE when SVM is taken as an algorithm that produces linear models
• Figure 4: $C-$index for expert feature selection, parametric RFE and automatic RFE when LR is taken as an algorithm that produces linear models