Understanding the characteristics of the housing market is essential for both sellers and buyers. However, the housing market is influenced by multiple factors. In this paper, the urban and socio-economic structure of an area is used to predict the price of 10387 properties sold in 2018 in the city of Dublin. More precisely, the direct distance from each property to 160 urban features taken from OpenStreetMap is calculated, and an extreme gradient boosting linear regression performed. Using these features, the model explains 45% of the housing price variance. The most important features in this model are the proximity to an embassy and to a grassland. In addition, the results of a population census from 2016 are also used to correlate with the price of properties. From this census, 48 features are used as the input of a gradient boost linear regression model. In all, the socio-economic features are explaining 43% of the housing price variance as well. The density of individuals reporting that they are not providing unpaid personal help for a friend or family member as well as individuals reporting that they have no religion are the most important socio-economic features. By taking into account either urban or socio-economic features, it is possible to accurately estimate housing prices and to predict their evolution.

Sharing personal information is an important way of communicating on social media. Among the information possibly shared, new sensors and tools allow people to share emotion information via facial emotion recognition. This paper questions whether people are prepared to share personal information such as their own emotion on social media. In the current study we examined how factors such as felt emotion, motivation for sharing on social media as well as personality affected participants' willingness to share self-reported emotion or facial expression online. By carrying out a Generalized Linear Mixed Model analysis, this study found that participants' willingness to share self-reported emotion and facial expressions was influenced by their personality traits and the motivation for sharing their emotion information that they were given. From our results we can conclude that the estimated level of privacy for certain emotional information, such as facial expression, is influenced by the motivation for sharing the information online.

Virtual Reality technologies present exciting opportunities for increasing the understanding of affect and emotion-related phenomena. However, they remain understudied both from the principles of data gathering and for their prospective use as annotation tools. A 2×2 mixed design experiment assessed the influence of presentation of emotion related social interactions in virtual reality and using more traditional database techniques. Participants rated the interactions along four emotion-related dimensions (empathy, rapport, trust and tone), for conversational “floor” dominance and how they made the raters “feel” on a valence dimension. Forty-three participants watched clips of a recorded conversation. Significant differences were found between the rapport, trust, emotional tone, floor dominance and subjective feeling ratings when comparing emotional valence of the materials; however, there was no effect of presentation medium for any measure-presenting in virtual reality made no difference to a raters annotations when compared to more traditional computer monitor style ratings. Implications for using virtual reality for database creation and annotation are discussed and the value is assessed given the current state of these technologies. We conclude that this does not diminish the need for research in Virtual Reality in Affective Computing but call for caution in the interpretation of results.

User eXperience studies with products, systems or services have significantly increased in companies in order to anticipate their commercial success. Among the user experience dimensions, emotions are predominant. However User eXperience studies used several concepts to refer to emotions and current measures still have some flaws. Consequently, this doctoral project aims firstly to provide a multi-componential approach of emotions based on a psychological view, and secondly to provide Affective Computing solutions in order to evaluate emotions in User eXperience studies. Through a study using hand-gesture interface devices, three components of users' emotions were simultaneously measured with self-reports: the subjective, cognitive and motivational components. The results point out the possibility of measuring different components in order to gain a better understanding of emotions triggered by products. They also point out that self-reports measures could be improved with Affective Computing solutions. In this perspective, two emotion assessment tools were developed: Oudjat and EmoLyse.