How Cross-Validation Can Go Wrong and What to Do About It

Neunhoeffer, Marcel; Sternberg, Sebastian

doi:10.1017/pan.2018.39

Cited by 31 publications

(20 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Using this comprehensive sample and a variety of techniques enables us to refine the conclusions of earlier studies. We complement recent assessments of machine learning methods in other fields, for instance, regarding the closely related task of predicting civil wars in political science (Neunhoeffer and Sternberg, 2018). We thereby seek to contribute to a realistic assessment of the strengths and limitations of the various methods, and to stimulate further research in this area.…”

Section: Introductionmentioning

confidence: 93%

An Evaluation of Early Warning Models for Systemic Banking Crises: Does Machine Learning Improve Predictions?

2018

View full text Add to dashboard Cite

Section: Introductionmentioning

confidence: 93%

An Evaluation of Early Warning Models for Systemic Banking Crises: Does Machine Learning Improve Predictions?

2018

View full text Add to dashboard Cite

“…For a specific time window size, the method collects all past features within the time window and current features as input to the learning algorithm. e time window used in this study is 5 ( Figure S1), which is selected by cross validation in the released set [33]. Figure 3 illustrates the scheme of our proposed method.…”

Section: Combining With Time Windowmentioning

confidence: 99%

Detection of Sleep Apnea from Single-Lead ECG Signal Using a Time Window Artificial Neural Network

Wang

Shen

2019

BioMed Research International

View full text Add to dashboard Cite

Sleep apnea (SA) is a ubiquitous sleep-related respiratory disease. It can occur hundreds of times at night, and its long-term occurrences can lead to some serious cardiovascular and neurological diseases. Polysomnography (PSG) is a commonly used diagnostic device for SA. But it requires suspected patients to sleep in the lab for one to two nights and records about 16 signals through expert monitoring. The complex processes hinder the widespread implementation of PSG in public health applications. Recently, some researchers have proposed using a single-lead ECG signal for SA detection. These methods are based on the hypothesis that the SA relies only on the current ECG signal segment. However, SA has time dependence; that is, the SA of the ECG segment at the previous moment has an impact on the current SA diagnosis. In this study, we develop a time window artificial neural network that can take advantage of the time dependence between ECG signal segments and does not require any prior assumptions about the distribution of training data. By verifying on a real ECG signal dataset, the performance of our method has been significantly improved compared to traditional non-time window machine learning methods as well as previous works.

show abstract

“…Onsets constitute 1.2% of cases. 18 I also report classical errors.19 The two coefficients are significantly different at p ¼ :049.20 Note thatNeunhoeffer & Sternberg (2019) cast doubt on the superior performance of random forests over penalized likelihood approaches.…”

mentioning

confidence: 99%

“… 20 Note that Neunhoeffer & Sternberg (2019) cast doubt on the superior performance of random forests over penalized likelihood approaches. …”

mentioning

confidence: 99%

International politics by other means: External sources of civil war

Toukan

2019

Journal of Peace Research

View full text Add to dashboard Cite

The literature on civil wars has recently turned towards their international context but lacks an account for how conflict beyond a state’s borders contributes to civil war onset. I argue that interstate rivalries can increase the risk of civil war in other states when rivals come to associate the foreign-policy orientation of other states with their own security. I present three pathways through which rivals increase the risk of civil war in other states. First, competition between rivals creates a ratchet effect by which the prospect of one’s involvement in a conflict makes it more likely that the other becomes involved. This dynamic makes support easier to secure and lowers the expected costs of war for governments and opposition groups. Second, rivals encourage domestic polarization as parties attempt to capture their influence, making domestic conflicts more intractable. Third, uncertainty over the potential for intervention by rivals increases the risk of miscalculation. I test the implications of the theory with novel spatial measures of interstate conflict and rivalry. Using logistic regressions and random forests, I find that being in the neighborhood of interstate rivals can increase a state’s risk of civil war.

show abstract

How Cross-Validation Can Go Wrong and What to Do About It

Cited by 31 publications

References 10 publications

An Evaluation of Early Warning Models for Systemic Banking Crises: Does Machine Learning Improve Predictions?

An Evaluation of Early Warning Models for Systemic Banking Crises: Does Machine Learning Improve Predictions?

Detection of Sleep Apnea from Single-Lead ECG Signal Using a Time Window Artificial Neural Network

International politics by other means: External sources of civil war

Contact Info

Product

Resources

About