東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

FindBook

Google Book

Amazon

博客來

Essays on Credit Swap Default Market : = Time Series Predictions with Machine Learning = Kredi Temerrut Piyasasi Uzerine Makaleler : Makine Ogrenmesi ile Zaman Serisi Tahminleri.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Essays on Credit Swap Default Market :/
其他題名:	Time Series Predictions with Machine Learning = Kredi Temerrut Piyasasi Uzerine Makaleler : Makine Ogrenmesi ile Zaman Serisi Tahminleri.
其他題名:	Kredi Temerrut Piyasasi Uzerine Makaleler :
作者:	Barokas, Lina.
面頁冊數:	1 online resource (110 pages)
附註:	Source: Masters Abstracts International, Volume: 84-04.
Contained By:	Masters Abstracts International84-04.
標題:	Neural networks. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29442849click for full text (PQDT)
ISBN:	9798352687703

Essays on Credit Swap Default Market : = Time Series Predictions with Machine Learning = Kredi Temerrut Piyasasi Uzerine Makaleler : Makine Ogrenmesi ile Zaman Serisi Tahminleri.
Barokas, Lina.

Essays on Credit Swap Default Market :Time Series Predictions with Machine Learning = Kredi Temerrut Piyasasi Uzerine Makaleler : Makine Ogrenmesi ile Zaman Serisi Tahminleri.Kredi Temerrut Piyasasi Uzerine Makaleler :Makine Ogrenmesi ile Zaman Serisi Tahminleri. - 1 online resource (110 pages)

Source: Masters Abstracts International, Volume: 84-04.

Thesis (Ph.D.)--Marmara Universitesi (Turkey), 2022.

Includes bibliographical references

in this dissertation, credit default swap risk premiums of BRiCS (Brazil, Russia, China, South Africa) and MiNT (Mexico, indonesia and Turkey) countries are predicted by applying the most recent ground-breaking techniques in machine learning. These techniques include Ridge regression, LASSO estimator and recurrent neural networks, which are Elman's RNN, NARX, GRU and LSTM. The predictive powers of these methods are compared by the mean absolute error, mean squared error and R-squared and these results differ by country and the type of the state-of-the-art forecaster. The contribution of each feature to the prediction is discussed by analyzing the SHAP values. The daily data is collected from 03/01/2012 and ends on 29/11/2019 to avoid COViD-19 effects on the state of the global economy. CDS risk premium for 5-year maturity contracts is used as an output variable and it is predicted by the selected local variables and global variables. in first chapter, in terms of higher R-squared obtained, LASSO outperforms Ridge regression in all selected countries. However, in terms of lower RMSE and the MAE scores, the lowest values are obtained by ridge regression for all selected countries. in second chapter, in all countries except China, the results of NARX outweigh Elman's RNN. in third chapter, GRU outperforms LSTM network for Brazil and indonesia. LSTM network outperforms GRU in Russia, South Africa, Mexico and Turkey. Both networks perform similarly for China. The results of Turkey model reached the highest forecast accuracy among the selected countries. Overall, recurrent neural networks outperform the ridge regression and LASSO estimator.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2023

Mode of access: World Wide Web

ISBN: 9798352687703Subjects--Topical Terms:

677449
Neural networks.
Index Terms--Genre/Form:

542853
Electronic books.

Essays on Credit Swap Default Market : = Time Series Predictions with Machine Learning = Kredi Temerrut Piyasasi Uzerine Makaleler : Makine Ogrenmesi ile Zaman Serisi Tahminleri.
LDR:04804nmm a2200361K 4500 001 2354798
005 20230501063923.5
006 m o d
007 cr mn ---uuuuu
008 241011s2022 xx obm 000 0 eng d
020 $a 9798352687703
035 $a (MiAaPQ)AAI29442849
035 $a (MiAaPQ)Marmara_11424280929
035 $a AAI29442849
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Barokas, Lina. $3 3695169
245 1 0 $a Essays on Credit Swap Default Market : $b Time Series Predictions with Machine Learning = Kredi Temerrut Piyasasi Uzerine Makaleler : Makine Ogrenmesi ile Zaman Serisi Tahminleri.
246 3 1 $a Kredi Temerrut Piyasasi Uzerine Makaleler : $b Makine Ogrenmesi ile Zaman Serisi Tahminleri.
264 0 $c 2022
300 $a 1 online resource (110 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Masters Abstracts International, Volume: 84-04.
500 $a Advisor: Celik, Sadullah.
502 $a Thesis (Ph.D.)--Marmara Universitesi (Turkey), 2022.
504 $a Includes bibliographical references
520 $a in this dissertation, credit default swap risk premiums of BRiCS (Brazil, Russia, China, South Africa) and MiNT (Mexico, indonesia and Turkey) countries are predicted by applying the most recent ground-breaking techniques in machine learning. These techniques include Ridge regression, LASSO estimator and recurrent neural networks, which are Elman's RNN, NARX, GRU and LSTM. The predictive powers of these methods are compared by the mean absolute error, mean squared error and R-squared and these results differ by country and the type of the state-of-the-art forecaster. The contribution of each feature to the prediction is discussed by analyzing the SHAP values. The daily data is collected from 03/01/2012 and ends on 29/11/2019 to avoid COViD-19 effects on the state of the global economy. CDS risk premium for 5-year maturity contracts is used as an output variable and it is predicted by the selected local variables and global variables. in first chapter, in terms of higher R-squared obtained, LASSO outperforms Ridge regression in all selected countries. However, in terms of lower RMSE and the MAE scores, the lowest values are obtained by ridge regression for all selected countries. in second chapter, in all countries except China, the results of NARX outweigh Elman's RNN. in third chapter, GRU outperforms LSTM network for Brazil and indonesia. LSTM network outperforms GRU in Russia, South Africa, Mexico and Turkey. Both networks perform similarly for China. The results of Turkey model reached the highest forecast accuracy among the selected countries. Overall, recurrent neural networks outperform the ridge regression and LASSO estimator.
520 $a Bu tezde, BRiCS (Brezilya, Rusya, Cin, Guney Afrika) ve MiNT (Meksika, Endonezya, Turkiye) ulkelerinin kredi temerrut riski, en yeni yontemler olan makine ogrenmesi ile tahmin edilmistir. Bu teknikler arasinda Ridge regresyonu, LASSO tahmin edicisi ve tekrarlayan sinir aglari (Elman RNN, NARX, LSTM, GRU) kullanilmistir. Kullanilan metotlarin tahmin kuvveti, ortalama mutlak hata, ortalama karekok hatasi ve anlamlilik katsayisi R ile karsilastirilmistir. Sonuclar ulke ve kullanilan metot bazinda farkliliklar gostermektedir. Kullanilan aciklayici degiskenlerin, hesaplanan tahminlere katkilari, SHAP degerleri analiz edilerek aciklanmistir. Calismada 03/01/2012'den 29/11/2019 tarihine kadar olan gunluk veri kullanilmistir. Tahmin edilen degisken, 5 yil vadeli kontratli kredi temerrut risk birimi olarak secilmistir. Ilk bolumun sonucunda, LASSO tahmin edicisi ile daha yuksek R katsayisi elde edilmistir. Ancak Ridge regresyonu ile daha dusuk ortalama mutlak hata ve ortalama karekok hatasi elde edilmistir. Ikinci bolumun sonucunda, Cin haricindeki ulkelerde, NARX tekrarlayan sinir agi, Elman tekrarlayan sinir agindan daha iyi sonuclar uretmistir. Son bolumde ise, Brezilya ve Endonezya icin GRU teknigi ile LSTM tekniginden daha iyi sonuclar elde edilmistir. Fakat LSTM, Rusya, Guney Afrika, Meksika ve Turkiye icin daha iyi sonuclar cikarmistir. Cin icin iki yonteminde sonuclari yakindir. Tahmin modellerinin performansini ulke bazli kiyaslarsak, sinir agi en yuksek tahmin kuvvetini Turkiye icin ulasmistir. Sonuc olarak, derin ogrenme yontemi olan tekrarlayan sinir aglari, makine ogrenmesi olan Ridge regresyonu ve LASSO tahmin edicisinden daha yuksek dogrulukta tahminler hesaplamistir.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2023
538 $a Mode of access: World Wide Web
650 4 $a Neural networks. $3 677449
655 7 $a Electronic books. $2 lcsh $3 542853
690 $a 0800
690 $a 0501
710 2 $a ProQuest Information and Learning Co. $3 783688
710 2 $a Marmara Universitesi (Turkey). $3 3555089
773 0 $t Masters Abstracts International $g 84-04.
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29442849 $z click for full text (PQDT)