In this thesis, we examine the machine learning models logistic regression, multilayer perceptron and random forests in the purpose of discriminate between good and bad credit applicants. In addition to these models we address the problem of imbalanced data with the Synthetic Minority Over-Sampling Technique (SMOTE). The data available have 273 286 entries and contains information about the invoice of the applicant and the credit decision process as well as information about the applicant. The data was collected during the period 2015-2017. With AUC-values at about 73%some patterns are found that can discriminate between customers that are likely to pay their invoice and customers that are not. However, the more advanced models only performed slightly better than the logistic regression.

Abstract not available/Sammanfattning ej tillgänglig.

In this thesis, the inverse distance weighting, different kriging methods, ordinary least squares and two variants of the geographically weighted regression was used to evaluate the spatial mapping abilities on an observed dataset and a simulated dataset. The two datasets contain the same bioclimatic variable, near-surface air temperature, uniformly distributed over the whole world. The observed dataset is the observed temperature of a global atmospheric reanalysis produced by ECMWF and the other being simulated temperature produced by SMHI’s climate model EC-earth 3.1. The data, initially containing space-time information during the time period 1993-2010 displayed no significant temporal variation when using a spatio-temporal variogram. However, each year displayed its own variation so each year was split where the different methods were used on the observed dataset to estimate a surface for each year that was then used to make comparisons to the simulated data. CLARA clustering was done on the observed geographical dataset in the hope to force the inverse distance weighting and the kriging methods to estimate a locally varying mean. However, the variograms produced displayed an irregular trend that would lead to inaccurate kriging weights. Geometric anisotropy variogram analysis was accounted for that displayed moderate anisotropy. Results show that the geographically weighted regression family outperformed the rest of the used methods in terms of root mean squared error, mean absolute error and bias. It was able to create a surface that had a high resemblance to the observed data.

No description available.

No description available.

Vi har genomfört en enkätundersökning om restaurangers och serveringars ekonomiska utsatthet och anpassning under coronapandemin. 139 restauranger och serveringar från Stockholms kommun valdes slumpmässigt ut för att delta i studien. Varje servering fick tillgång till ett formulär, som bland annat inkluderade frågor relaterade till beställnings- och överlämningsmetoder samt försäljning. Svaren från respondenterna användes sedan för att göra punkt- och intervallskattningar om olika parametrar i populationen. Vi tar även fram en logistisk regressionsmodell för att skatta sannolikheter för att serveringar genomför olika anpassningsmetoder. Utifrån urvalet kan vi se att många serveringar har valt att införa någon sorts ny strategi som svar på pandemin. För de flesta har dock försäljningen minskat. Det verkar också som att serveringar som har fått färre kunder tenderar att ha en högre sannolikhet att anpassa sig. På grund av att bara runt 15,83 procent av de kontaktade serveringarna valde att svara, finns det stor osäkerhet i våra skattningar. / We have conducted a survey about the adaptation and economic vulnerability of restaurants and food services during the COVID-19 pandemic. 139 restaurants from Stockholm Municipality were randomly chosen to participate. Each restaurant was provided with a form that included questions related to different methods for ordering and serving food, and sales. The answers from the respondents were later used to form point and interval estimates for different population parameters. We have also used a logistic regression model to estimate the probability that food services utilize a new adaptation strategy. The results from the sample indicate that many food services have chosen some kind of new strategy in response to the pandemic. For most, sales have decreased. It also seems to be the case that food services that have had a decrease in customers tend to have higher probability for adaptation. Since only 15.83 percent of the food services that were included in the sample chose to participate, there exists large uncertainty in our estimates.

No description available.

No description available.

Syftet med denna uppsats var att med hjälp av Markovkedjor göra en statistisk prognos för ensnöfylld jul för tre svenska väderstationer. Datamängderna som ligger till grund förskattningarna av övergångssanolikheterna sorterades så endast perioden 17 december – 31december för varje år användes. Fundamentet för prognosen bygger på den underliggandeteorin, framförallt av den så kallade Markovegenskapen och Chapman-Kolmogorovsekvation. Prognosen har två utgångspunkter: antingen är det snö den 17 december eller inte.Uppsatsen undersöker också kedjornas ergodicitet och kedjornas stationära fördelningar -experimentellt, och med en teoretisk lösning. Sannolikheten för snö på julafton ansågs somgoda, förutsatt att det är snö den 17 dec. Att tillämpa Markovkedjor för denna sortens problemfungerade bra, vilket beror på det beroende som finns mellan dagens väder och vädretimorgon. / The purpose of this paper was to make a statistical forecast for a snow-filled Christmas forthree Swedish weather stations with the help of Markov chains. The data on which thetransition probabilities estimates were based on, were sorted so that only the period December17 – December 31 for each year was used. The basis for the forecast is based on theunderlying theory, the so-called Markov property and the Chapman-Kolmogorov equation.The forecast has two starting points: whether it is snow on December 17 or not. The thesisalso examined the ergodicity and the stationary distributions of the Markov chains –experimentally and with a theoretical solution. The probability of snow on Christmas Eve wasconsidered good, provided there is snow on Dec 17. Applying Markov chains to this kind ofproblem worked well, due to the dependency that exists between today’s weather and theweather tomorrow

This paper investigates which factors best predict the economic state of a Swedish municipality after the 2008 crisis by constructing a linear model that regresses the change in the unemployment rate on a set of variables. The variables used for the model were from a dataset put together using data from a government service and were selected for the model using Bayesian information criterion. From this procedure, a model with six independent variables was estimated. The model’s statistics were examined, and the model was subsequentially tried against the five multiple linear regression assumptions. It was concluded that the model did not fulfil the assumption of homoscedasticity, and because of this, the dependent variable was transformed into a logarithm, thus yielding a log-lin model. This model ended up fulfilling every assumption and had higher explanatory power than the previous model. It is concluded that the variables that denote the number of newly registered businesses per 1000 residents, the share of residents with a high education, the fraction of net-commuters, the number of refugees received with a residence permit per 1000 residents, total net investments per person, the share of long term unemployed residents and the population size all prove significant when included together in a log-lin model of the change in the unemployment rate.