Abstract
The road transportation sector is responsible for 87% of the human CO2 emissions. The estimation and prediction of fuel consumption plays a key role in the development of systems that foster the reduction of those emissions through trip planing. In this paper, we present a predictive regression model of instantaneous fuel consumption for diesel and gasoline light-duty vehicles, based on their instantaneous speed and acceleration and on road inclination. The parameters are extracted from GPS data, thus the models do not require data from dedicated vehicle sensors. We use data collected by 17 drivers during their daily commutes using the SenseMyCity crowdsensor. We perform an empyrical comparison of several regression algorithms for prediction across trips of the same vehicle and for prediction across vehicles. The results show that models trained for a vehicle show similar RMSE when are applied to other vehicles with similar characteristics. Relying on these results, we propose fuel type specific models that provide an accurate prediction for vehicles with similar characteristics to those on which the models were trained.

Abstract
The large amount of Recommender System algorithms makes the selection of the most suitable algorithm for a new dataset a difficult task. Metalearning has been successfully used to deal with this problem. It works by mapping dataset characteristics with the predictive performance obtained by a set of algorithms. The models built on this data are capable of predicting the best algorithm for a new dataset. However, typical approaches try only to predict the best algorithm, overlooking the performance of others. This study focus on the use of Metalearning to select the best ranking of CF algorithms for a new recommendation dataset. The contribution lies in the formalization and experimental validation of using Label Ranking to select a ranked list of algorithms. The experimental procedure proves the superior performance of the proposed approach regarding both ranking accuracy and impact on the baselevel performance. Furthermore, it draws and compares the knowledge regarding metafeature importance for both classification and Label Ranking tasks in order to provide guidelines for the design of algorithms in the Recommender System community.

Abstract
Exceptional preferences mining (EPM) is a crossover between two subfields of data mining: local pattern mining and preference learning. EPM can be seen as a local pattern mining task that finds subsets of observations where some preference relations between labels significantly deviate from the norm. It is a variant of subgroup discovery, with rankings of labels as the target concept. We employ several quality measures that highlight subgroups featuring exceptional preferences, where the focus of what constitutes exceptional' varies with the quality measure: two measures look for exceptional overall ranking behavior, one measure indicates whether a particular label stands out from the rest, and a fourth measure highlights subgroups with unusual pairwise label ranking behavior. We explore a few datasets and compare with existing techniques. The results confirm that the new task EPM can deliver interesting knowledge.

Abstract
Food trucks are a widely popular fast food restaurant alternative, whose differentiating factor is their proximity to customers. Their popularity has stimulated the expansion of available options, which now includes several different types of cuisines, consequently making the choice by customers a challenging issue. From data obtained via a market research, in which hundreds of participants provided their food truck preferences, this paper focuses on the problem of food truck recommendation using a multilabel approach. In particular, it investigates how to improve the recommendation task regarding a previous work, where some labels have never been predicted. In order to address this problem, different alternatives were investigated. One of these alternatives, the Ensemble of Single Label, proposed in this paper, was able to reduce it. Despite its simplicity, good predictive results were obtained when they were used in the investigated task. Among other benefits, all labels were correctly predicted at least for few instances.

Abstract
The algorithm selection problem refers to the ability to predict the best algorithms for a new problem. This task has been often addressed by Metalearning, which looks for a function able to map problem characteristics to the performance of a set of algorithms. In the context of Collaborative Filtering, a few studies have proposed and validated the merits of different types of problem characteristics for this problem (i.e. dataset-based approach): using systematic metafeatures and performance estimations obtained by subsampling landmarkers. More recently, the problem was tackled using Collaborative Filtering models in a novel framework named CF4CF. This framework leverages the performance estimations as ratings in order to select the best algorithms without using any data characteristics (i.e algorithm-based approach). Given the good results obtained independently using each approach, this paper starts with the hypothesis that the integration of both approaches in a unified algorithm selection framework can improve the predictive performance. Hence, this work introduces CF4CF-META, an hybrid framework which leverages both data and algorithm ratings within a modified Label Ranking model. Furthermore, it takes advantage of CF4CF’s internal mechanism to use samples of data at prediction time, which has proven to be effective. This work starts by explaining and formalizing state of the art Collaborative Filtering algorithm selection frameworks (Metalearning, CF4CF and CF4CF-META) and assess their performance via an empirical study. The results show CF4CF-META is able to consistently outperform all other frameworks with statistically significant differences in terms of meta-accuracy and requires fewer landmarkers to do so. © 2018, Springer Nature Switzerland AG.

Abstract
As Collaborative Filtering becomes increasingly important in both academia and industry recommendation solutions, it also becomes imperative to study the algorithm selection task in this domain. This problem aims at inding automatic solutions which enable the selection of the best algorithms for a new problem, without performing full-ledged training and validation procedures. Existing work in this area includes several approaches using Metalearning, which relate the characteristics of the problem domain with the performance of the algorithms. This study explores an alternative approach to deal with this problem. Since, in essence, the algorithm selection problem is a recommendation problem, we investigate the use of Collaborative Filtering algorithms to select Collaborative Filtering algorithms. The proposed approach integrates subsampling landmarkers, a data characterization approach commonly used in Metalearning, with a Collaborative Filtering methodology, named CF4CF. The predictive performance obtained by CF4CF using benchmark recommendation datasets was similar or superior to that obtained with Metalearning.