In the past decade, machine learning (ML) models have become farmore powerful, and are increasingly being used in many important contexts. At the same time, ML models have become more complex, and harder to understand on their own, which has necessitated an interesting explainable AI (XAI), a field concerned with ensuring that ML and other AI system can be understood by human users and practitioners. One aspect of XAI is the development of ”explainers”, tools that take a more complex system (here: an ML model) and generate a simpler but sufficiently accurate model of this system — either globally or locally —to yield insight into the behaviour of the original system. As ML models have become more complex and prevalent, concerns that they may embody and perpetuate harmful social biases have also risen, with XAI being one proposed tool for bias detection. This paper investigates the ability of two explainers, LIME and SHAP, which explain the prediction of potentially more complex models by way of locally faithful linear models, to detect harmful social bias (here in the form of the influence of the racial makeup of a neighbourhood on property values), in a simple experiment involving two kinds of ML models, line arregression and an ensemble method, trained on the well-known Boston-housing dataset. The results show that LIME and SHAP appear to be helpful in bias detection, while also revealing an instance where the explanations do not quite reflect the workings of the model, while still yielding accurate insight into the predictions the model makes.

Churn prediction methods are widely used in Customer Relationship Management and have proven to be valuable for retaining customers. To obtain a high predictive performance, recent studies rely on increasingly complex machine learning methods, such as ensemble or hybrid models. However, the more complex a model is, the more difficult it becomes to understand how decisions are actually made. Previous studies on machine learning interpretability have used a global perspective for understanding black-box models. This study explores the use of local explanation models for explaining the individual predictions of a Random Forest ensemble model. The churn prediction was studied on the users of Tink – a finance app. This thesis aims to take local explanations one step further by making comparisons between churn indicators of different user groups. Three sets of groups were created based on differences in three user features. The importance scores of all globally found churn indicators were then computed for each group with the help of local explanation models. The results showed that the groups did not have any significant differences regarding the globally most important churn indicators. Instead, differences were found for globally less important churn indicators, concerning the type of information that users stored in the app. In addition to comparing churn indicators between user groups, the result of this study was a well-performing Random Forest ensemble model with the ability of explaining the reason behind churn predictions for individual users. The model proved to be significantly better than a number of simpler models, with an average AUC of 0.93. / Metoder för att prediktera utträde är vanliga inom Customer Relationship Management och har visat sig vara värdefulla när det kommer till att behålla kunder. För att kunna prediktera utträde med så hög säkerhet som möjligt har den senasteforskningen fokuserat på alltmer komplexa maskininlärningsmodeller, såsom ensembler och hybridmodeller. En konsekvens av att ha alltmer komplexa modellerär dock att det blir svårare och svårare att förstå hur en viss modell har kommitfram till ett visst beslut. Tidigare studier inom maskininlärningsinterpretering har haft ett globalt perspektiv för att förklara svårförståeliga modeller. Denna studieutforskar lokala förklaringsmodeller för att förklara individuella beslut av en ensemblemodell känd som 'Random Forest'. Prediktionen av utträde studeras påanvändarna av Tink – en finansapp. Syftet med denna studie är att ta lokala förklaringsmodeller ett steg längre genomatt göra jämförelser av indikatorer för utträde mellan olika användargrupper. Totalt undersöktes tre par av grupper som påvisade skillnader i tre olika variabler. Sedan användes lokala förklaringsmodeller till att beräkna hur viktiga alla globaltfunna indikatorer för utträde var för respektive grupp. Resultaten visade att detinte fanns några signifikanta skillnader mellan grupperna gällande huvudindikatorerna för utträde. Istället visade resultaten skillnader i mindre viktiga indikatorer som hade att göra med den typ av information som lagras av användarna i appen. Förutom att undersöka skillnader i indikatorer för utträde resulterade dennastudie i en välfungerande modell för att prediktera utträde med förmågan attförklara individuella beslut. Random Forest-modellen visade sig vara signifikantbättre än ett antal enklare modeller, med ett AUC-värde på 0.93.

Yes / This paper examines the impacts of clay mineralogy on the effectiveness of lime stabilisation at different temperatures. A comprehensive experimental programme was conducted to track down the evolution of lime-clay reactions and their durations through monitoring the evolution of strength gain at predetermined times using the Unconfined Compressive Strength (UCS) test. The study examined clays with different mineralogy compositions comprising Na+ Bentonite and Ball (Kaolinite) clay. Four different clays were tested including 100% bentonite, 100% Ball clay and two clay mixtures with ratios of 1:1 and 1:3 by mass of bentonite to Ball clay. All clays were treated using a range of lime content up to 25% and cured for a period of time up to 672 h at two different temperatures of 20 and 40 °C. The results showed that the continuity of the fast phase (stage 1) of strength gain was dependent on the availability of lime in particular at the higher temperature. Whereas, for the same lime content, the duration of the fast phase and the kinetic of strength gain were significantly related to the clay mineralogy and curing temperature. Except for the initial strength gain at 0 h curing time, the lime-treated Ball clay specimens at 20 °C appeared to show no strength gain throughout the curing period that extended up to 672 h. However, when curing occurred at 40 °C, the no strength gain stage only lasted for 72 h after which a gradual increase in the strength was observed over the remaining curing period of time. The addition of Bentonite to Ball clay succeeded in kicking off the strength gain after a short period of curing time at both curing temperatures.

Natural soils in work-sites are sometimes detrimental to the construction of engineering projects. Problematic soils such as soft and expansive soils are a real source of concern to the long-term stability of structures if care is not taken. Expansive soils could generate immense distress due to their volume change in response to a slight change in their water content. On the other hand, soft soils are characterised by their low shear strength and poor workability. In earthwork, replacing these soils is sometimes economically and sustainably unjustifiable in particular if they can be stabilised to improve their behaviour. Several techniques have evolved to enable construction on problematic soils such as reinforcement using fibre and planar layers and piled reinforced embankments. Chemical treatment using, e.g. lime and/or cement is an alternative method to seize the volume change of swelling clays. The use of lime as a binding agent is becoming a popular method due to its abundant availability and cost-effectiveness. When mixed with swelling clays, lime enhances the mechanical properties, workability and reduces sensitivity to absorption and release of water. There is a consensus in the literature about the primary mechanisms, namely cation exchange, flocculation and pozzolanic reaction, which cause the changes in the soil characteristics after adding lime in the presence of water. The dispute is about whether these mechanisms occur in a sequential or synchronous manner. More precisely, the controversy concerns the formation of cementitious compounds in the pozzolanic reaction, whether it starts directly or after the cation exchange and flocculation are completed. The current study aims to monitor the signs of the formation of such compounds using a geotechnical approach. In this context, the effect of delayed compaction, lime content, mineralogy composition, curing time and environmental temperature on the properties of lime-treated clays were investigated. The compaction, swelling and permeability, and unconfind compression strength tests were chosen to evaluate such effect. In general, the results of the geotechnical approach have been characterised by their scattering. The sources of this dispersion are numerous and include sampling methods, pulverisation degree, mixing times and delay of compaction process, a pre-test temperature and humidity, differences in dry unit weight values, and testing methods. Therefore, in the current study, several precautions have been set to reduce the scattering in the results of such tests so that they can be used efficiently to monitor the evolution in the properties that are directly related to the formation and development of cementitious compounds. Four clays with different mineralogy compositions, covering a wide range of liquid limits, were chosen. The mechanical and hydraulic behaviour of such clays that had been treated by various concentrations of lime up to 25% at two ambient temperatures of 20 and 40oC were monitored for various curing times. The results indicated that the timing of the onset of changes in mechanical and hydraulic properties that are related to the formation of cementitious compounds depends on the mineralogy composition of treated clay and ambient temperature. Moreover, at a given temperature, the continuity of such changes in the characteristics of a given lime-treated clay depends on the lime availability.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.