|
|
Nowcasting the Real GDP Growth of the European Economies based on Machine Learning
Baylan, Su Hazal ; Kočenda, Evžen (advisor) ; Baruník, Jozef (referee)
This thesis analyzes the nowcasting of quarterly GDP growth for nine European economies using a dynamic factor model and four different machine learning models. These machine learning models are as follows: Ridge, Lasso, Elastic Net, and Random Forest. The data includes ten hard and fifteen soft indicators for each country in order to calculate GDP for each nowcasting iteration for pre-covid and covid periods. For machine learning, models are fed with the extracted factors that are obtained from the dynamic factor model, and for all nowcasting models expanding window approach is selected to estimate nowcasting iterations. The empirical finding indicates that overall machine learning models provide better forecasting accuracy compared to dynamic factor models and benchmark models for more stable periods, such as the period before Covid-19. On the other hand, for more volatile periods where the uncertainties are higher in economies, the dynamic factor model outperforms machine learning models in order to nowcast GDP growth. In addition to this, Random Forest is able to outperform all the alternative models for small economies such as Slovenia and Portugal for stable periods. JEL Classification C01, C33, C53, C83, E37 Keywords Nowcasting, DFM, Ridge, Lasso, Elastic Net, Random Forest Title Nowcasting...
|
|
|
Sparse Representation of Signals
Mesárošová, Michaela ; Arm, Jakub (referee) ; Jirgl, Miroslav (advisor)
People who are immobile or lie for long periods are at high risk of developing pressure ulcers and require additional care. Therefore, it is necessary to monitor the condition of such persons as simply and efficiently as possible. In this work, we focus on processing the signals provided by a pressure mattress with a 30x11 sensor grid on which a person lays and the possibilities of its use after conversion into sparse representation coefficients. Redundant dictionaries, also known as frames, enable non-orthogonal representation of signals, which leads to a sparse representation of coefficients. Since this approach provides many advantageous properties and is being used in various applications, such as denoising, segmentation, robust transformations, quantum theory, and others, we verified the possibility of classifying a person’s lying position based on a sparse representation. The results were compared with other traditional classification methods, which were found to be less suitable for the classification problem, with the best-achieved result of 92.41 % for CNN, but with high demands on time, design and complexity. The success rate of the classification reached 92.76 %, with fewer demands on design and implementation complexity. The possibilities of classification and reconstruction of an image containing occlusions were also investigated, where the sparse representation proved to be an effective method to remove these defects.
|
|
|
Forecasting realized volatility using machine learning and mixed-frequency data (the case of the Russian stock market)
Pyrlik, Vladimir ; Elizarov, P. ; Leonova, A.
We assess the performance of selected machine learning algorithms (lasso, random forest, gradient boosting, and long short-term memory) in forecasting the daily realized volatility of returns of selected top stocks in the Russian stock market in comparison with a heterogeneous autoregressive realized volatility benchmark in 2018-2020. We seek to improve the predictive power of the models by including various economic indicators that carry information about future volatility. We find that lasso delivers a good combination of easy implementation and forecast precision. The other algorithms require fine-tuning and frequent re-training, otherwise they are likely to fail to outperform the benchmark often enough. Only the basic lagged log-RV values are significant explanatory variables in terms of the benchmark in-sample quality. Many economic indicators of mixed frequencies improve the predictive power of lasso though, including calendar and overnight effects, financial spillovers from local and global markets, and various macroeconomics indicators.
|
|
|
Variable selection based on penalized likelihood
Chlubnová, Tereza ; Kulich, Michal (advisor) ; Maciak, Matúš (referee)
Selection of variables and estimation of regression coefficients in datasets with the number of variables exceeding the number of observations consti- tutes an often discussed topic in modern statistics. Today the maximum penalized likelihood method with an appropriately selected function of the parameter as the penalty is used for solving this problem. The penalty should evaluate the benefit of the variable and possibly mitigate or nullify the re- spective regression coefficient. The SCAD and LASSO penalty functions are popular for their ability to choose appropriate regressors and at the same time estimate the parameters in a model. This thesis presents an overview of up to date results in the area of characteristics of estimates obtained by using these two methods for both small number of regressors and multidimensional datasets in a normal linear model. Due to the fact that the amount of pe- nalty and therefore also the choice of the model is heavily influenced by the tuning parameter, this thesis further discusses its selection. The behavior of the LASSO and SCAD penalty functions for different values and possibili- ties for selection of the tuning parameter is tested with various numbers of regressors on simulated datasets.
|
|
|
Comparison of different models for forecasting of Czech electricity market
Kunc, Vladimír ; Krištoufek, Ladislav (advisor) ; Kopečná, Vědunka (referee)
There is a demand for decision support tools that can model the electricity markets and allows to forecast the hourly electricity price. Many different ap- proach such as artificial neural network or support vector regression are used in the literature. This thesis provides comparison of several different estima- tors under one settings using available data from Czech electricity market. The resulting comparison of over 5000 different estimators led to a selection of several best performing models. The role of historical weather data (temper- ature, dew point and humidity) is also assesed within the comparison and it was found that while the inclusion of weather data might lead to overfitting, it is beneficial under the right circumstances. The best performing approach was the Lasso regression estimated using modified Lars. 1
|
|
|
Post-selection Inference: Lasso & Group Lasso
Bouř, Vojtěch ; Maciak, Matúš (advisor) ; Kulich, Michal (referee)
The lasso is a popular tool that can be used for variable selection and esti- mation, however, classical statistical inference cannot be applied for its estimates. In this thesis the classical and the group lasso is described together with effici- ent algorithms for the solution. The key part is dedicated to the post-selection inference for the lasso estimates where we explain why the classical inference is not suitable. Three post-selection tests for the lasso are described and one test is proposed also for the group lasso. The tests are compared in simulations where finite sample properties are examined. The tests are further applied on a practical example. 1
|
|
|
Variable selection based on penalized likelihood
Chlubnová, Tereza ; Kulich, Michal (advisor) ; Maciak, Matúš (referee)
Selection of variables and estimation of regression coefficients in datasets with the number of variables exceeding the number of observations consti- tutes an often discussed topic in modern statistics. Today the maximum penalized likelihood method with an appropriately selected function of the parameter as the penalty is used for solving this problem. The penalty should evaluate the benefit of the variable and possibly mitigate or nullify the re- spective regression coefficient. The SCAD and LASSO penalty functions are popular for their ability to choose appropriate regressors and at the same time estimate the parameters in a model. This thesis presents an overview of up to date results in the area of characteristics of estimates obtained by using these two methods for both small number of regressors and multidimensional datasets in a normal linear model. Due to the fact that the amount of pe- nalty and therefore also the choice of the model is heavily influenced by the tuning parameter, this thesis further discusses its selection. The behavior of the LASSO and SCAD penalty functions for different values and possibili- ties for selection of the tuning parameter is tested with various numbers of regressors on simulated datasets.
|
|
|
The Lasso and its application to time series
Holý, Vladimír ; Prášková, Zuzana (advisor) ; Hendrych, Radek (referee)
This thesis first describes the Lasso method and its adaptive improvement. Then the basic theoretical properties are shown and different algorithms are introduced. The main part of this thesis is application of the Lasso method to AR, MA and ARCH time series and to REGAR, REGMA and REGARCH models. An algorithm of the adaptive Lasso in a more general time series model, which includes all above mentioned models and series, is developed. The properties of methods and algorithms are shown on simulations and on a practical example. Powered by TCPDF (www.tcpdf.org)
|
|
|
Analýza regionálních cen nemovitostí ve Spojených státech pomocí vysokodimenzionálního VAR modelu
Krčál, Adam ; Čížek, Ondřej (advisor) ; Zouhar, Jan (referee)
In this thesis the heterogeneity of regional real estate prices in United States is investigated. A high dimensional VAR model with additional exogenous predictors, originally introduced by \cite{fan11}, is adopted. In this framework, the common factor in regional house prices dynamics is explained by exogenous predictors and the spatial dependencies are captured by lagged house prices in other regions. For the purpose of estimation and variable selection under high-dimensional setting the concept of Penalized Least Squares (PLS) with different penalty functions (e.g. LASSO penalty) is studied in detail and implemented. Moreover, clustering methods are employed to identify subsets of statistical regions with similar house prices dynamics. It is demonstrated that these clusters are well geographically defined and contribute to a better interpretation of the VAR model. Next, we make use of the LASSO variable selection property in order to construct the impulse response functions and to simulate the prices behavior when a shock occurs. And last but not least, one-period-ahead forecasts from VAR model are compared to those from the Diffusion Index Factor Model by \cite{stock02}, a commonly used model for forecasts.
|
guest ::
