Syftet med denna uppsats är att modellera och prognostisera Sveriges varuexport, varuimport och handelsnetto. Vi använder oss av data från januari 1975 till och med augusti 2010 för respektive serie. Dessa data testas och jämförs i olika ARIMA- och SARIMA-modeller samt skattas även medelst säsongsreningsprogrammet TRAMO/SEATS. För de modeller som bäst passar serierna genomförs därefter in sample- och out of sample-analyser med felmåtten RMSE och MAPE. Modellerna med bäst felmått och som därpå väljs ut för att göra prognoser för serierna till och med augusti 2012 är (3,1,0)x(0,1,1) för export, (2,1,1)x(0,1,1) för import samt (0,1,1)x(0,1,1) skattad i TRAMO/SEATS för handelsnetto.

The Swedish Transport Agency has for a long time collected data on a monthly basis for different variables that are used to make predictions, short projections as well as longer projections. They have used SAS for producing statistical models in air transport. The model with the largest value of coefficient of determination is the method that has been used for a long time. The Swedish Transport Agency felt it was time for an evaluation of their models and methods of how projections is estimated, they would also explore the possibilities to use different, completely new models for forecasting air travel. This Bachelor thesis examines how the Holt-Winters method does compare with SARIMA, error terms such as RMSE, MAPE, R2, AIC and BIC  will be compared between the methods.  The results which have been produced showing that there may be a risk that the Holt-Winters models adepts a bit too well in a few variables in which Holt-Winters method has been adapted. But overall the Holt-Winters method generates better forecasts . / <p>Avbryt / Spara utkast</p>

Made available in DSpace on 2014-06-12T17:16:37Z (GMT). No. of bitstreams: 2 arquivo2907_1.pdf: 634979 bytes, checksum: 330e453c0db3f5452e436a3247c47be0 (MD5) license.txt: 1748 bytes, checksum: 8a4605be74aa9ea9d79846c1fba20a33 (MD5) Previous issue date: 2009 / Esta dissertação tem como principal objetivo apresentar os modelos de previsão de arrecadação do ICMS, por segmento econômico, para a Secretaria da Fazenda do Estado de Pernambuco, utilizando as técnicas econométricas. Objetiva-se, com essa pesquisa, disponibilizar aos gestores púbicos do Estado mais um modelo de previsão consistente e com certo grau de confiabilidade. Para tanto, utilizou-se da metodologia Box-Jenkins, mais especificamente os modelos: ARIMA - modelo autorregressivo integrado de média móvel, e SARIMA - modelo autorregressivo integrado de média móvel sazonal, e o software RATS (Regression Analyse Time Series). O trabalho apresenta o comportamento da arrecadação de ICMS no Estado e uma revisão da literatura, onde são abordados os principais conceitos teóricos utilizados, bem como uma análise dos resultados obtidos. Conclui-se que o modelo de previsão utilizando séries temporais, em função de sua capacidade preditiva, pode se transformar em um valioso instrumento para auxiliar na elevação da receita tributária no Estado de Pernambuco, dentro da capacidade contributiva de cada contribuinte

Aujourd'hui, La Poste se trouve dans une situation particulièrement délicate au regard des mutations de son environnement économique. Pour répondre à ses nouveaux enjeux, elle doit développer sa planification stratégique, dans laquelle la prévision de son chiffre d'affaires joue un rôle particulièrement crucial. Or, à l'heure actuelle, les méthodes utilisées par la Direction Stratégique, notamment pour traiter la question de l'effet jours ouvrables, ne sont pas optimales et l'entreprise cherche à les améliorer. Notre thèse, réalisée en convention CIFRE avec la Direction Marketing Stratégique de La Poste, s'inscrit dans ce questionnement. Notre recherche vise plus spécifiquement à déterminer quels sont les meilleurs modèles économétriques pour la prévision du chiffre d'affaires du courrier. On se penche dans un premier temps sur la question de l'effet jours ouvrables que l'on traite à l'aide de méthodes de prévision, afin d'en obtenir une analyse approfondie. Puis on cherche à déterminer des modèles de prévisions adaptés à chaque type de clientèle et, enfin, au chiffre d'affaires totales. Pour l'entreprise, cette recherche vise à élaborer un outil fiable de prévision et d'aide à la décision. Au point de vue théorique, le principal apport de notre travail réside dans l'utilisation de modèles de prévision pour analyser l'effet jours ouvrables, à la place de l'utilisation d'outils de détection automatique. / Nowadays, La Poste is facing a particularly complex situation, related to the many changes of its economic environment. In order to respond to the new issues, it must develop strategic planning, in which income prediction plays a crucial part. Yet, to this day, the methods used by the Strategy Department are not optimal and the company is working at their improvement. Our research,conducted in the framework of a CIFRE partnership with the Strategic Marketing Department in La Poste, is anchored in this questioning. Our work is specifically aimed at determining the best econometric models to predict income of the Mail activity. We first focus on the issue of the "Trading days effect", that we examine using prediction methods, in order to get an in-depth view of it. Then we engage in determining prediction models adapted to each type of customers and, finally, a model for total income. For the company, this research is aimed at elaborating a reliable prediction and decision-making tool. From the theoretical point of view, the main contribution of our work lies in our using prediction models to analyze "Trading days effect", instead of automatic detection tools.

No description available.

Enligt det antagna klimatmålet år 2017 ska Sverige vara utsläppsneutralt år 2045. Detta innebär att Sveriges utsläpp av växthusgaser måste minska kraftigt. I denna studie studeras utsläpp från olika sektorer och genom en tidsserieanalys prediceras framtida utsläppsvärden fram. Resultaten i studien visar att utsläppen förväntas minska inom samtliga studerade sektorer men att ytterligare forskning krävs för att veta om denna minskning är tillräcklig. Vidare är resultaten känsliga för förändringar, varför uppdaterade och kontinuerliga analyser rekommenderas.

The present dissertation deals with an analysis of monthly time series of average temperatures and precipitation sums recorded at 44 sites in the Czech Republic over the period of 1961--2008. The main research purpose is to acquire deeper knowledge of regularities in the climatic time series development, using an appropriate set of statistical methods. A secondary objective is to search and find correlations between the research outcomes and basic geographic coordinates (altitude, longitude and latitude) of particular measurement stations and comparing all the results achieved for the selected climatic elements. There are two major contributions of this work. In the first place, it presents new knowledge in the field of climatic time series, particularly in connection with the strength and development of their seasonal component, further for instance analysing the relation between the distribution of a residual component and the geographic coordinates of the measurement stations. Another contribution lies in an extensive application of statistical methods of climatic time series analysis. Several types of methods were used, having employed both widely and rarely applied statistical tools (linear trends analysis and Box-Jenkins methodology respectively) as well as those used for the very first time (moving-seasonal time series).

Submitted by Maria Medeiros (maria.dilva1@ufcg.edu.br) on 2018-08-13T15:28:50Z No. of bitstreams: 1 POLLYANNA KELLY DE OLIVEIRA SILVA - TESE (PPGMet) 2017.pdf: 11004478 bytes, checksum: 0d5e098181f432beffc2fd8155027f1e (MD5) / Made available in DSpace on 2018-08-13T15:28:50Z (GMT). No. of bitstreams: 1 POLLYANNA KELLY DE OLIVEIRA SILVA - TESE (PPGMet) 2017.pdf: 11004478 bytes, checksum: 0d5e098181f432beffc2fd8155027f1e (MD5) Previous issue date: 2017-08-30 / CNPq / O vento como fonte para geração de energia elétrica é analisado neste trabalho através de sua variabilidade e da obtenção de previsões de curto prazo para o ano de 2010, período de atuação de El Niño-Oscilação Sul (ENOS) moderado. Modelos de séries temporais propostos por Box-Jenkins e o indicador de desempenho de predição MMREE são usados para obter as melhores estimativas da velocidade do vento com base nas séries observadas. São utilizados dados anemométricos do Projeto SONDA situado às margens do Rio São Francisco em Petrolina – PE, e de dois parques eólicos localizados no litoral do Estado do Ceará: Quixaba (litoral leste), na cidade de Aracati, e Lagoa Seca (litoral oeste), na cidade de Acaraú. O ciclo diário do vento tem velocidades mais baixas (altas) no período da madrugada-início da manhã (pela manhã e final da noite, com exceção do litoral oeste, cujas máximas ocorrem no final da tarde). Um cisalhamento vertical negativo, no vento local, é observado em períodos distintos do dia nas três áreas de estudo. No Ceará ele ocorre no período da manhã (início da tarde e meio da noite) no litoral leste (oeste) e no Lago de Sobradinho durante a noite até o início da manhã. Foi observado que no litoral leste os ventos são mais fortes, provavelmente devido à curvatura côncava do litoral. As estimativas da velocidade do vento no horizonte de 24 horas pelo modelo SARIMA, com dados horários dos 30 dias anteriores ao dia da previsão para treino (Caso 2), mostraram redução nos erros e melhora significativa na série estimada no período da madrugada-início da manhã; no Lago de Sobradinho essas estimativas são mais precisas, quando comparadas àquelas feitas com base em toda a série de dados (Caso 1). Os resultados indicam que o modelo SARIMA com período de entrada de dados menor pode ser aplicado para a previsão da velocidade do vento em áreas de potencial eólico, dando suporte ao operador da rede elétrica na programação da geração despachável para o dia seguinte. / The wind as a source for power generation is analyzed in this work by means of its variability and short-range wind forecasts for the year of 2010, period of moderate El Niño-Southern Oscillation (ENSO). Time series models proposed by Box-Jenkins and the indicator of forecast accuracy MMREE are used to obtain the best wind speed estimates based on the observed series. Anemometric data of the SONDA Project located on the shore of the São Francisco River in Petrolina-PE, and of two wind power plants located on the coast of the Ceará State, Quixaba (east coast), in the city of Aracati, and Lagoa Seca (west coast), in the city of Acaraú, are used. The daily wind cycle has lower (higher) speeds in late night-early morning (in the morning and end of the night, with exception of the west coast, whose maxima occur in late afternoon). A negative vertical shear in the local wind is observed in distinct periods of the day in the three study areas. In Ceará it occurs in the morning (early afternoon and middle of the night) on the east (west) coast and on Sobradinho Lake at night until early in the morning. It was observed that the winds are stronger on the east coast, probably due to the coast’s concave curvature. The wind speed estimates in a 24-hour horizon by the SARIMA model, with hourly data of the 30 days that precede the forecast day for training (Case 2), showed reduction in the errors and significant improvement in the estimated series in late night-early morning; in Sobradinho Lake these estimates are more accurate, as compared to the estimates based on the entire data series (Case 1). The results indicate that the SARIMA model with horter time series as input may be applied to forecast wind speed in areas of eolic potential, giving support to the system operator in programming the dispatchable distributed generation for the next day.

In this Master Thesis there are summarized basic methods for modelling time series, such as linear regression with seasonal dummy variables, exponential smoothing and SARIMA processes. The thesis is aimed on modelling and forecasting seasonal time series using these methods. Goals of the Thesis are to introduce and compare these methods using a set of 2184 seasonal time series followed by evaluation their prediction abilities. The main benefit of this Master Thesis is understanding of different aspects of forecasting time series and empirical verification of advantages and disadvantages these methods in field of creating predictions.

Huvudsyftet med denna studie är att förstå om ett säsongsmässigt autoregressivt integrerat rörligt genomsnitt (SARIMA) -metod pålitligt kan förutsäga extrem variation i värmelaster för en fjärrvärmestation. Genom extrem variation ser vi på den maximala och minsta värmebelastningen per dag mätt i megawattstimmar. Avhandlingen bygger på standardimplementering av SARIMAX och utför en rutnätsökning efter de mest lämpliga parametrarna. Prognoser kan genereras från tidsserier i syfte att uppskatta förväntad energiförbrukning i en fjärrvärmestation. Frågan som ställs är: Hur tillförlitlig är SARIMAX-modellen för energibehov i en fjärrvärmestation? För att besvara studiens fråga designas och genomförs experiment med hjälp av ett dataset från verkliga mätningar. Datasetet studerades och analyserades med hjälp av undersökande dataanalystekniker som kommer med statistiska paket implementerade i en pythonmiljön, som kan användas som ett statistiskt program. Uppgifterna är uppdelade i två säsonger, sommar och vinter. Där den explorativa analysen av datasetet visar att modellen måste ta hänsyn till den starka veckocykeln med data. Så att korrelationen mellan utetemperaturen kan användas för att förbättra förutsägelsen. Fininställning och tillämpning av SARIMAX och Prophet för förutsägelser genererar data i form av diagram som visar hur tillförlitlig modellen är för förutsägelse. Resultaten visar att SARIMAX-modellen presterar bättre under vintermånaderna och sämre under sommaren. Baserat på dessa resultat antyder avhandlingsstudien att SARIMAX-modellen är mer tillämplig under vintermånaderna där förutsägelsen är mer tillförlitlig. Jämförelser med Prophet modellen indikerar lovande resultat och att vidare forskning borde föras för denna modell. Dessa resultat kan vara till hjälp för industrin som förser samhället och konsumenterna med fjärrvärme. Det hjälper till att förutse hur mycket energiförbrukning som används där industrin kan använda den för att reglera mängden fjärrvärme, för att ytterligare hjälpa ekonomin och miljön. / The main objective in this study is to understand if a Seasonal Autoregressive Integrated Moving Average (SARIMA) method can reliably predict extreme variation in heat loads for a district heating substation. By extreme variation we look at the maximum and minimum heat load per day measured in megawatt hour. The thesis relies on standard implementation of SARIMAX and performs a grid search for the most suitable parameters. Forecast can be generated from time series with the purpose of estimating expected energy consumption in a district heating substation. The question addressed is: How reliable is the SARIMAX-model for energy demand in a district heating substation? To answer the study’s question, experiments are designed and conducted using a dataset from real measurements. The dataset was studied and analyzed using exploratory data analysis techniques that come with statistical packages implemented in the python environment, which can be used as a statistical program. The data is separated into two seasons, summer and winter. Where the explorative analysis of the data shows that the model needs to take in account the strong weekly cycle of data. Also the correlation between the outside temperature can be used to improve prediction. Fine tuning and applying SARIMAX and Prophet for predictions generates data in the form of graphs and tables which shows how reliable the SARIMAX model is for prediction. Results show that the SARIMAX model is performing better during winter months and worse during summer. Based on these results, the thesis study suggests that the SARIMAX-model is more applicable during winter months where prediction is more reliable. Comparison with the Prophet model indicates promising results and that further investigations should be made into this model. These results can be of help to the industry that supplies the community and consumers with district heating. It helps by predicting how much energy consumption is used where the industry can use it to regulate the amount of district heating, to further help the economy and environment.