Browsing by Subject "prediction"

Now showing 1 - 13 of 13

Access status: Open Access ,
A new methodology that surpasses current bridging theories to efficiently seal a varied pore throat distribution as found in natural reservoir formations
(2006) Vickers, Stephen; Cowie, Martin; Jones, Thomas A.; Twynam, Allan J.
This paper presents a new method (model) for predicting the most efficient way to create a particle bridge on a formation face having a wide spread of pore throat sizes. Laboratory data, based on particle bridging distributions using this new model, are shown to be an improvement over two current industry methods. The improvement using this design has been effective in the field and has resulted high productivity.
Access status: Restricted ,
Analysis of computer network state prediction based on machine learning
(Data obrony: 2019-12-09) Śliwiński, Miłosz
Wydział Informatyki, Elektroniki i Telekomunikacji
Access status: Open Access ,
Energetic raw material prices in the aspect of defining mutual interactions and relations
(Wydawnictwa AGH, 2005) Ranosz, Robert; Kopacz, Michał
Energetic raw materials markets have played, and still do, a fundamental role in defining the worldwide economic relations. It is especially important in the field of world energy economy and energy safety. As far as the analysis of the situation in this area that we have at our disposal shows and allows understanding the current trends and foreseeing the future market tendencies, the globalisation increasingly brings about concentration of production in a centre of the worldwide business. Let's then venture a statement that essential connections exist between the fundamentals of the raw material economy (production, production capacities, consumption, proved reserves) in the context of modelling of economical parameters, particularly prices.
Access status: Open Access ,
Geofizyka w inżynierii i ochronie środowiska
(Wydawnictwa AGH, 2009) Zuberek, Wacław M.
The main goals of geophysics have been considered in following areas: - prediction and prevention of natural catastrophes and mitigation of their impact, - detection, location and monitoring of contaminated areas and goals related to their reclamation, - goals related to the safety of large and important structures and constructions, - goals related to safe deposition, storage and monitoring of dangerous wastes and toxic materials, - application of geophysical methods and observations to obtain »clean« and renewable energy, - detection, location and dating of archeological materials and monuments of cultural heritage. Due to the great importance of prediction in environmental engineering and protection, its possibilities and limitations have been considered separately.
Access status: Restricted ,
Link prediction based on classical techniques and dual convolutional neural networks
(Data obrony: 2019-07-12) Borkowski, Grzegorz
Wydział Informatyki, Elektroniki i Telekomunikacji
Access status: Restricted ,
Modelowanie i predykcja zmienności czasowej produkcji biogazu w wybranych komunalnych oczyszczalniach ścieków
(Data obrony: 2014-10-08) Mirski, Bartosz
Wydział Geologii, Geofizyki i Ochrony Środowiska
The aim of presented thesis is modeling and prediction of temporal variability of biogas in the three selected municipal wastewater treatment plants in Krosno, Rzeszow and Krakow. The data were prepared and modeled in Statistica's Statsoft program on the basis of materials obtained from municipal wastewater plants. Modeling was done by using stepwise regression and neural networks. Next stage was prediction of data. The accuracy of prediction was estimated by four measures: MAPE, Theil coefficient, Janus coefficient and RMSPE. Similar models could be used in municipal wastewater treatment plants to estimate the amount of biogas production. It will be used to product heat energy and electrical power which consequently is to reduce the wastewater treatment plants requirements for the outside energy intake.
Access status: Restricted ,
Optymalizacja giełdowej strategii inwestycyjnej za pomocą algorytmu genetycznego
(Data obrony: 2019-01-21) Domarecki, Sebastian
Wydział Elektrotechniki, Automatyki, Informatyki i Inżynierii Biomedycznej
Access status: Open Access ,
Predictions and Application of Queueing Analysis: Case of Regional Hospital Limbe, Cameroon
(Wydawnictwa AGH, 2024) Machangara, Daphne T.; Boubacar, Habiboulaye Amadou; Andreatta, Giovanni; Ndolo, Antony
In this work, we applied queue analysis and the predictions of waiting times at Regional Hospital Limbe (RHL) in Cameroon. The main purpose of the work was to be able to make mathematical sense of a real-life scenario that concerned queues (waiting lines) and try to come up with models for performance measurements and improvements; this could be achieved by using queueing theory concepts that were composed of queueing models that provided some operational insights because of their analytical nature. The observations included studying patient arrival and waiting times, along with doctor service times; the results showed busy departments in the hospital, busy days, and busy times. Long waiting times were mainly found to exist in general practitioner (GP) and specialist consultations. The queueing concept was applied to only one service segment – GP consultation. Although strong scientific conclusions cannot be made on the queuing models that were obtained due to inefficient data, the value of this work lies mainly in the methodologies and proposals of different operating systems that could be adopted. Furthermore, some predictions were made using machine learning to see how long a patient could wait in a queue for service; the model predictions had an average of 10 minutes and 53 seconds of error.
Access status: Restricted ,
Statystyczna analiza pól losowych określonych na sferze i elipsoidzie
(Data obrony: 2011-10-21) Balawejder-Czaja, Katarzyna
Wydział Matematyki Stosowanej
Access status: Open Access ,
The new approach to estimation of the hazard function in business demography on example of data from New Zealand
(2013) Zając, Paweł
The author presents the new methodology for the estimation of the hazard function for the new born enterprises', survival rate called FIRM. The methodology is based on construction of a stochastic process and is examined in the Monte Carlo simulation study with real data. The dataset is provided by Statistics New Zealand and contains all enterprises born in period between 2001-2010. Enterprises are divided in clusters according to the number of employees and for each cluster individual simulations are made. Achieved coefficients of determination in clusters are around 90%. The author finds substantial differences in survival probability according to employee count size in the company. Simulations done in this study allow to estimate mean and standard deviation of life duration for enterprises and prediction of the hazard function for each cluster.
Access status: Restricted ,
Uzupełnianie brakujących lub błędnych fragmentów profilowań geofizyki otworowej na podstawie innych pomiarów z wykorzystaniem sztucznych sieci neuronowych
(Data obrony: 2019-01-25) Piwko, Aleksandra
Wydział Geologii, Geofizyki i Ochrony Środowiska
Na podstawie zestawu kompletnych profilowań geofizyki otworowej w części otworu Ja-2 skonstruowano i przeprowadzono proces uczenia sieci neuronowej metodą wstecznej propagacji błędu w celu estymacji indeksu absorpcji elektrycznej (Pe). Następnie zaaplikowano utworzoną sieć do niekompletnych danych w dalszej części otworu i obliczono rozkład parametru (Pe) z głębokością w otworze Ja-2. W oparciu o dane stratygraficzne wybrano w otworze Ja-3 interwał głębokościowy o podobnej litologii. Przy pomocy wcześniej utworzonego algorytmu obliczono rozkład tego samego parametru w wybranym fragmencie otworu Ja-3.
Access status: Open Access ,
Wpływ filtracji cyfrowej na predykcję cen i zapasów miedzi
(Wydawnictwa AGH, 2007) Świder, Mariusz
Artykuł ten opisuje możliwości poprawienia predykcji notowań cen i zasobów miedzi przy użyciu filtracji cyfrowej. Stanowi ona etap wstępny, który eliminuje z sygnału losowego, jakim jest notowanie giełdowe, jego wyższe harmoniczne traktowane jako zakłócenia. Po analizie widmowej ciągu notowań giełdowych określono najwłaściwsze częstotliwości próbkowania i wybrano dwa najbardziej odpowiednie rodzaje filtrów cyfrowych. Dla tak przefiltrowanych sygnałów oraz dla notowań oryginalnych użyto metody predykcji GARCH. Jest ona szeroko stosowana w modelowaniu ekonometrycznym i finansowym, ze względu na dobre wyniki symulacji zachowań procesów heteroskedastycznych, jakimi są notowania giełdowe. Na końcu porównano błędy predykcji, uzyskując zadowalająco niskie ich wartości dla pierwszych 200 próbek sygnałów odfiltrowanych, które można porównać z błędem predykcji sygnału oryginalnego.
Access status: Open Access ,
Wykorzystanie podejścia immunologicznego do prognozowania szeregów czasowych
(Wydawnictwa AGH, 2009) Pełech-Pilichowski, Tomasz; Duda, Jan Tadeusz
This article presents an idea of application of immune paradigm to symptomatic event detection, announcing long-term trend changes in time series that give a possibility to improve efficiency of short and medium-term prediction based on linear trend extrapolation. A conception of immune-like algorithm and definition of main system objects are presented, exemplary results of calculation (applied to financial time series) given with partially implementation of proposed immune-based algorithm are showed.