Short-Term Forecasting of Well Production Based on a Hybrid Probabilistic Approach

The presented article is dedicated to creation and testing of probabilistic ensemble computational tool for operational forecasting of well production in short term (STF). The ensemble consisted of models based on such physical and mathematical tools as: the equation of non-stationary filtration, material balance, Darcy's law and machine learning models. After calculations by each model, their forecasts are combined into a single ensemble forecast. The hybrid approach is based on the Monte Carlo method on Markov chains as a separate probabilistic model using Bayes’ formula. In this case, statistical weights of each model (the degree of confidence in each model) is determined in the form of a probability distribution based on the reliability of previously performed forecasts. The test results presented in this article were obtained on the real field data. The obtained forecasts of individual models and the ensemble were compared to real data. Real data tool usage analysis showed that the proposed approach gives a small error in comparison with actual measurements. Efficiency of calculations allows to automatically adapt the model to the entire well production history (several hundred wells) within a few hours.

Combining Results of Performance-Based and Informant Test Accuracy Studies: Bayes or Boole?

<b><i>Background/Aims:</i></b> Since screening and diagnostic tests for dementia do not have perfect accuracy, &#x3e;1 test is often administered when assessing patients with cognitive complaints. Use of both patient performance tests and informant questionnaires has been recommended. Combination of individual test results may be based on methods originally defined by Thomas Bayes (revision or updating of pretest probabilities to post-test probabilities given the test results) and by George Boole (application of associative “AND” or “OR” operator). This study sought to apply these methods in clinical practice. <b><i>Methods:</i></b> Using the dataset of a pragmatic test accuracy study of the Six-Item Cognitive Impairment Test (6CIT) and informant Ascertain Dementia 8 (AD8), post-test probabilities for the combination were calculated using Bayes’ formula and compared to Boolean “AND” combination. Combined test sensitivity and specificity was calculated using either Boolean “AND” or “OR” operator and compared to results using equations based on individual test sensitivity and specificity. <b><i>Results:</i></b> Both Bayesian and Boolean methods produced similar improvements from pretest probability (0.288) to combined post-test probability for dementia (≈0.5). Likewise, the 2 different methods for calculating combined sensitivities and specificities gave similar results, with, as anticipated, the “AND” combination improving overall specificity (to ≈0.65) whereas the “OR” combination improved sensitivity (to ≈1.00). <b><i>Conclusion:</i></b> Combination of individual screening test results using Bayesian and Boolean methods is relatively straightforward and may add to clinicians’ intuitive judgements when combining test results.

Diagnostic evaluation and Bayesian Updating: Practical solutions to common problems

This article discusses several practical issues arising with the application of diagnostic principles to theory-based evaluation (e.g. with Process Tracing and Bayesian Updating). It is structured around three iterative application steps, focusing mostly on the third. While covering different ways evaluators fall victims to confirmation bias and conservatism, the article includes suggestions on which theories can be tested, what kind of empirical material can act as evidence and how to estimate the Bayes formula values/update confidence, including when working with ranges and qualitative confidence descriptors. The article tackles evidence packages (one of the most problematical practical issues), proposing ways to (a) set boundaries of single observations that can be considered independent and handled numerically; (b) handle evidence packages when numerical probability estimates are not available. Some concepts are exemplified using a policy influence process where an institution’s strategy has been influenced by a knowledge product by another organisation.

Integrals and integral transformations related to the vector Gaussian distribution

This paper is dedicated to the integrals and integral transformations related to the probability density function of the vector Gaussian distribution and arising in probability applications. Herein, we present three integrals that permit to calculate the moments of the multivariate Gaussian distribution. Moreover, the total probability formula and Bayes formula for the vector Gaussian distribution are given. The obtained results are proven. The deduction of the integrals is performed on the basis of the Gauss elimination method. The total probability formula and Bayes formula are obtained on the basis of the proven integrals. These integrals and integral transformations could be used, for example, in the statistical decision theory, particularly, in the dual control theory, and as table integrals in various areas of research. On the basis of the obtained results, Bayesian estimations of the coefficients of the multiple regression function are calculated.

Implementation of Integrated Bayes Formula and Support Vector Machine for Analysing Airline’s Passengers Review

Nowadays, the utilization of Internet of Things (IoT) is commonly used in the tourism industry, including aviation, where passengers of flight services can rate their satisfaction levels towards the product and service they use by writing their reviews in the form of text-based data on many popular websites. These passenger reviews are collections of potential big data and can be analyzed in order to extract meaningful informations. Some text mining algorithms are already in common use, including the Bayes formula and Support Vector Machine methods. This research proposes an implementation of the Bayes and SVM methods where these algorithms will operate independently yet integrated with other modules such as input data, text pre-processing and shows output result concisely in one single information system. The proposed system was successfully delivered 1000 documents of passenger reviews as input data, then after implemented the pre-processing method, the Bayes formula was used to classify the document reviews into 5 categories, including plane condition, flight comfort, staff service, food and entertainment, and price. While simultanously, the positive and negative sentiment contained in the review document was analyzed with SVM method and shows the accuracy score of 83.6% for a training to testing set ratio of 50:50, while 82.75% accuracy for the 60:40 ratio, and 83.3% accuracy for the 70:30 ratio. This research shows that two different text mining algorithms can be implemented simultaneously in a effective and efficient way, while still providing an accurate and satisfying performance results in one integrated information system.

A69 Phylodynamics of language evolution

We used Bayesian evolutionary analysis to study linguistic data and infer phylogenetic trees of language evolution. Languages were encoded as binary strings indicating the presence or absence of members of cognate classes, the equivalence of classes of words with similar meaning, and shared ancestry. These strings formed the alignment data used to compute the posterior likelihood of a tree with respect to Bayes’ formula. Informative priors are crucial for testing hypotheses regarding the age of common ancestry and divergence times and should include as much available information as possible. Here, we investigated the birth–death process as a method to construct tree priors specifically suitable for modeling the evolution of cognate data. To test these models, we will use a dataset of the languages from Vanuatu, an island nation featuring world’s highest language density.

PROBABILISTIC INFERENCE FOR INTERVAL PROBABILITIES IN DECISION-MAKING PROCESSES

The present paper considers one approach to Bayes’ formula based probabilistic inference under interval values of relevant probabilities; the necessity of it is caused by the impossibility to obtain reliable deterministic values of the required probabilistic evaluations. The paper shows that the approach proves to be the best from the viewpoint of the required amount of calculations and visual representation of the results. The execution of the algorithm of probabilistic inference is illustrated using a classical task of decision making related to oil mining. For visualisation purposes, the state of initial and target information is modelled using probability trees.