scholarly journals Self–Training With Quantile Errors for Multivariate Missing Data Imputation for Regression Problems in Electronic Medical Records: Algorithm Development Study (Preprint)

2021 ◽  
Author(s):  
Hansle Gwon ◽  
Imjin Ahn ◽  
Yunha Kim ◽  
Hee Jun Kang ◽  
Hyeram Seo ◽  
...  
Keyword(s):  
Machine Learning ◽  
Incomplete Data ◽  
Missing Values ◽  
Pearson Correlation ◽  
Laboratory Data ◽  
Complete Data ◽  
Learning System ◽  
Training Data ◽  
Rank Test ◽  
Missing Value

BACKGROUND When using machine learning in the real world, the missing value problem is the first problem encountered. Methods to impute this missing value include statistical methods such as mean, expectation-maximization, and multiple imputations by chained equations (MICE) as well as machine learning methods such as multilayer perceptron, k-nearest neighbor, and decision tree. OBJECTIVE The objective of this study was to impute numeric medical data such as physical data and laboratory data. We aimed to effectively impute data using a progressive method called self-training in the medical field where training data are scarce. METHODS In this paper, we propose a self-training method that gradually increases the available data. Models trained with complete data predict the missing values in incomplete data. Among the incomplete data, the data in which the missing value is validly predicted are incorporated into the complete data. Using the predicted value as the actual value is called pseudolabeling. This process is repeated until the condition is satisfied. The most important part of this process is how to evaluate the accuracy of pseudolabels. They can be evaluated by observing the effect of the pseudolabeled data on the performance of the model. RESULTS In self-training using random forest (RF), mean squared error was up to 12% lower than pure RF, and the Pearson correlation coefficient was 0.1% higher. This difference was confirmed statistically. In the Friedman test performed on MICE and RF, self-training showed a <i>P</i> value between .003 and .02. A Wilcoxon signed-rank test performed on the mean imputation showed the lowest possible <i>P</i> value, 3.05e-5, in all situations. CONCLUSIONS Self-training showed significant results in comparing the predicted values and actual values, but it needs to be verified in an actual machine learning system. And self-training has the potential to improve performance according to the pseudolabel evaluation method, which will be the main subject of our future research.

scholarly journals Self–Training With Quantile Errors for Multivariate Missing Data Imputation for Regression Problems in Electronic Medical Records: Algorithm Development Study

10.2196/30824 ◽  
2021 ◽  
Vol 7 (10) ◽  
pp. e30824
Author(s):  
Hansle Gwon ◽  
Imjin Ahn ◽  
Yunha Kim ◽  
Hee Jun Kang ◽  
Hyeram Seo ◽  
...  
Keyword(s):  
Machine Learning ◽  
Incomplete Data ◽  
Missing Values ◽  
Pearson Correlation ◽  
Laboratory Data ◽  
Complete Data ◽  
Learning System ◽  
Rank Test ◽  
P Value ◽  
Missing Value

Background When using machine learning in the real world, the missing value problem is the first problem encountered. Methods to impute this missing value include statistical methods such as mean, expectation-maximization, and multiple imputations by chained equations (MICE) as well as machine learning methods such as multilayer perceptron, k-nearest neighbor, and decision tree. Objective The objective of this study was to impute numeric medical data such as physical data and laboratory data. We aimed to effectively impute data using a progressive method called self-training in the medical field where training data are scarce. Methods In this paper, we propose a self-training method that gradually increases the available data. Models trained with complete data predict the missing values in incomplete data. Among the incomplete data, the data in which the missing value is validly predicted are incorporated into the complete data. Using the predicted value as the actual value is called pseudolabeling. This process is repeated until the condition is satisfied. The most important part of this process is how to evaluate the accuracy of pseudolabels. They can be evaluated by observing the effect of the pseudolabeled data on the performance of the model. Results In self-training using random forest (RF), mean squared error was up to 12% lower than pure RF, and the Pearson correlation coefficient was 0.1% higher. This difference was confirmed statistically. In the Friedman test performed on MICE and RF, self-training showed a P value between .003 and .02. A Wilcoxon signed-rank test performed on the mean imputation showed the lowest possible P value, 3.05e-5, in all situations. Conclusions Self-training showed significant results in comparing the predicted values and actual values, but it needs to be verified in an actual machine learning system. And self-training has the potential to improve performance according to the pseudolabel evaluation method, which will be the main subject of our future research.

scholarly journals Comparison of Algorithms for Clustering Incomplete Data

2014 ◽  
Vol 39 (2) ◽  
pp. 107-127 ◽  
Author(s):  
Artur Matyja ◽  
Krzysztof Siminski
Keyword(s):  
Data Analysis ◽  
Incomplete Data ◽  
Missing Values ◽  
Real Data ◽  
Complete Data ◽  
The Other ◽  
Data Sets ◽  
Missing Value ◽  
Comparison Of Algorithms ◽  
New Algorithms

Abstract The missing values are not uncommon in real data sets. The algorithms and methods used for the data analysis of complete data sets cannot always be applied to missing value data. In order to use the existing methods for complete data, the missing value data sets are preprocessed. The other solution to this problem is creation of new algorithms dedicated to missing value data sets. The objective of our research is to compare the preprocessing techniques and specialised algorithms and to find their most advantageous usage.

scholarly journals Evolutionary Machine Learning for Classification with Incomplete Data

2021 ◽  
Author(s):  
◽  
Cao Truong Tran
Keyword(s):  
Machine Learning ◽  
Feature Selection ◽  
Genetic Programming ◽  
Incomplete Data ◽  
Missing Values ◽  
Machine Learning Techniques ◽  
Feature Construction ◽  
Classification Algorithms ◽  
Learning Techniques ◽  
Effectiveness And Efficiency

<p>Classification is a major task in machine learning and data mining. Many real-world datasets suffer from the unavoidable issue of missing values. Classification with incomplete data has to be carefully handled because inadequate treatment of missing values will cause large classification errors.    Existing most researchers working on classification with incomplete data focused on improving the effectiveness, but did not adequately address the issue of the efficiency of applying the classifiers to classify unseen instances, which is much more important than the act of creating classifiers. A common approach to classification with incomplete data is to use imputation methods to replace missing values with plausible values before building classifiers and classifying unseen instances. This approach provides complete data which can be then used by any classification algorithm, but sophisticated imputation methods are usually computationally intensive, especially for the application process of classification. Another approach to classification with incomplete data is to build a classifier that can directly work with missing values. This approach does not require time for estimating missing values, but it often generates inaccurate and complex classifiers when faced with numerous missing values. A recent approach to classification with incomplete data which also avoids estimating missing values is to build a set of classifiers which then is used to select applicable classifiers for classifying unseen instances. However, this approach is also often inaccurate and takes a long time to find applicable classifiers when faced with numerous missing values.   The overall goal of the thesis is to simultaneously improve the effectiveness and efficiency of classification with incomplete data by using evolutionary machine learning techniques for feature selection, clustering, ensemble learning, feature construction and constructing classifiers.   The thesis develops approaches for improving imputation for classification with incomplete data by integrating clustering and feature selection with imputation. The approaches improve both the effectiveness and the efficiency of using imputation for classification with incomplete data.   The thesis develops wrapper-based feature selection methods to improve input space for classification algorithms that are able to work directly with incomplete data. The methods not only improve the classification accuracy, but also reduce the complexity of classifiers able to work directly with incomplete data.   The thesis develops a feature construction method to improve input space for classification algorithms with incomplete data by proposing interval genetic programming-genetic programming with a set of interval functions. The method improves the classification accuracy and reduces the complexity of classifiers.   The thesis develops an ensemble approach to classification with incomplete data by integrating imputation, feature selection, and ensemble learning. The results show that the approach is more accurate, and faster than previous common methods for classification with incomplete data.   The thesis develops interval genetic programming to directly evolve classifiers for incomplete data. The results show that classifiers generated by interval genetic programming can be more effective and efficient than classifiers generated the combination of imputation and traditional genetic programming. Interval genetic programming is also more effective than common classification algorithms able to work directly with incomplete data.    In summary, the thesis develops a range of approaches for simultaneously improving the effectiveness and efficiency of classification with incomplete data by using a range of evolutionary machine learning techniques.</p>

scholarly journals Predicting Fault Slip via Transfer Learning

2021 ◽  
Author(s):  
Kun Wang ◽  
Christopher Johnson ◽  
Kane Bennett ◽  
Paul Johnson
Keyword(s):  
Machine Learning ◽  
Numerical Simulations ◽  
Transfer Learning ◽  
Laboratory Experiments ◽  
Laboratory Data ◽  
Fault Slip ◽  
Geophysical Data ◽  
Training Data ◽  
Data Sets ◽  
Earthquake Cycle

Abstract Data-driven machine-learning for predicting instantaneous and future fault-slip in laboratory experiments has recently progressed markedly due to large training data sets. In Earth however, earthquake interevent times range from 10's-100's of years and geophysical data typically exist for only a portion of an earthquake cycle. Sparse data presents a serious challenge to training machine learning models. Here we describe a transfer learning approach using numerical simulations to train a convolutional encoder-decoder that predicts fault-slip behavior in laboratory experiments. The model learns a mapping between acoustic emission histories and fault-slip from numerical simulations, and generalizes to produce accurate results using laboratory data. Notably slip-predictions markedly improve using the simulation-data trained-model and training the latent space using a portion of a single laboratory earthquake-cycle. The transfer learning results elucidate the potential of using models trained on numerical simulations and fine-tuned with small geophysical data sets for potential applications to faults in Earth.

scholarly journals Effectiveness, Explainability and Reliability of Machine Meta-Learning Methods for Predicting Mortality in Patients with COVID-19: Results of the Brazilian COVID-19 Registry

2021 ◽  
Author(s):  
Bruno Barbosa Miranda de Paiva ◽  
Polianna Delfino Pereira ◽  
Claudio Moises Valiense de Andrade ◽  
Virginia Mara Reis Gomes ◽  
Maria Clara Pontello Barbosa Lima ◽  
...  
Keyword(s):  
Machine Learning ◽  
Prediction Models ◽  
State Of The Art ◽  
Laboratory Data ◽  
Machine Learning Algorithms ◽  
Training Data ◽  
Learning Models ◽  
Learning Methods ◽  
Meta Learning ◽  
Machine Learning Models

Objective: To provide a thorough comparative study among state ofthe art machine learning methods and statistical methods for determining in-hospital mortality in COVID 19 patients using data upon hospital admission; to study the reliability of the predictions of the most effective methods by correlating the probability of the outcome and the accuracy of the methods; to investigate how explainable are the predictions produced by the most effective methods. Materials and Methods: De-identified data were obtained from COVID 19 positive patients in 36 participating hospitals, from March 1 to September 30, 2020. Demographic, comorbidity, clinical presentation and laboratory data were used as training data to develop COVID 19 mortality prediction models. Multiple machine learning and traditional statistics models were trained on this prediction task using a folded cross validation procedure, from which we assessed performance and interpretability metrics. Results: The Stacking of machine learning models improved over the previous state of the art results by more than 26% in predicting the class of interest (death), achieving 87.1% of AUROC and macroF1 of 73.9%. We also show that some machine learning models can be very interpretable and reliable, yielding more accurate predictions while providing a good explanation for the why. Conclusion: The best results were obtained using the meta learning ensemble model Stacking. State of the art explainability techniques such as SHAP values can be used to draw useful insights into the patterns learned by machine-learning algorithms. Machine learning models can be more explainable than traditional statistics models while also yielding highly reliable predictions. Key words: COVID-19; prognosis; prediction model; machine learning

scholarly journals Missing Value Imputation Approach for Mass Spectrometry-based Metabolomics Data

2017 ◽  
Author(s):  
Runmin Wei ◽  
Jingye Wang ◽  
Mingming Su ◽  
Erik Jia ◽  
Tianlu Chen ◽  
...  
Keyword(s):  
Mass Spectrometry ◽  
Missing Values ◽  
Pearson Correlation ◽  
Imputation Accuracy ◽  
Metabolomics Data ◽  
Missing Value ◽  
Sample Distribution ◽  
Imputation Methods ◽  
Missing Value Imputation ◽  
Squared Error

AbstractIntroductionMissing values exist widely in mass-spectrometry (MS) based metabolomics data. Various methods have been applied for handling missing values, but the selection of methods can significantly affect following data analyses and interpretations. According to the definition, there are three types of missing values, missing completely at random (MCAR), missing at random (MAR), and missing not at random (MNAR).ObjectivesThe aim of this study was to comprehensively compare common imputation methods for different types of missing values using two separate metabolomics data sets (977 and 198 serum samples respectively) to propose a strategy to deal with missing values in metabolomics studies.MethodsImputation methods included zero, half minimum (HM), mean, median, random forest (RF), singular value decomposition (SVD), k-nearest neighbors (kNN), and quantile regression imputation of left-censored data (QRILC). Normalized root mean squared error (NRMSE) and NRMSE-based sum of ranks (SOR) were applied to evaluate the imputation accuracy for MCAR/MAR and MNAR correspondingly. Principal component analysis (PCA)/partial least squares (PLS)-Procrustes sum of squared error were used to evaluate the overall sample distribution. Student’s t-test followed by Pearson correlation analysis was conducted to evaluate the effect of imputation on univariate statistical analysis.ResultsOur findings demonstrated that RF imputation performed the best for MCAR/MAR and QRILC was the favored one for MNAR.ConclusionCombining with “modified 80% rule”, we proposed a comprehensive strategy and developed a public-accessible web-tool for missing value imputation in metabolomics data.

scholarly journals Missing by Design Patterns for Optimizing Survey Response by Efficient and Consistent Data Collection

2020 ◽  
Author(s):  
◽  
Sara Bahrami
Keyword(s):  
Optimal Design ◽  
Design Patterns ◽  
Missing Values ◽  
Complete Data ◽  
Training Data ◽  
Lower Probability ◽  
Analysis Model ◽  
Vast Number ◽  
Ex Post ◽  
Highly Correlated

Respondent burden due to long questionnaires in surveys can negatively affect the response rate as well as the quality of responses. A solution to this problem is to use split questionnaire design (SQD). In an SQD, the items of the long questionnaire are divided into subsets and only a fraction of item-subsets are assigned to random subsamples of individuals. This will lead to several shorter questionnaires which are administered to random subsample of individuals. The completed sub-questionnaires are then combined and the missing values due to design are imputed by means of multiple imputation method. Identification problems can be avoided in advance by ensuring that the combination of variables in the analysis model of interest are jointly observed on at least a subsample of individuals. Furthermore, including an appropriate combination of items in each sub-questionnaire is the most important concern in designing the SQD to reduce the information loss, i.e. highly correlated items that explain each other well should not be jointly missing. For this reason, training data must be available from previous surveys or a pilot study to exploit the association between the variables. In this thesis two SQDs are proposed. In the first study a potential design for NEPS data is introduced. The data consist of items which can be divided and allocated into blocks according to their context, with the objective that the within block correlations are higher relative to the between block correlations. According to the design, the target sample is divided to subsamples. In addition to the items of a whole block which is assigned to each subsample, a fraction of items of the remaining blocks are randomly drawn and assigned to each subsample. Where items that belong to blocks with relatively higher correlations are drawn with lower probability. The design is evaluated by means of several ex-post investigations. The design is imposed on complete data and several models are estimated for both complete data and data deleted by design. The design is also compared with a random multiple matrix sampling design which assigns random subset of items to each sample individual. In the second study, a genetic algorithm is used to search among a vast number of SQDs to find the optimal design. The algorithm evaluates the designs by the fraction of missing information (FMI) induced by the design. The optimal design is the one with the smallest FMI. The optimal design is evaluated by means of several simulation studies and is compared with a random MMS design.

scholarly journals Automatic missing value imputation for cleaning phase of diabetic’s readmission prediction model

2022 ◽  
Vol 12 (2) ◽  
pp. 2001
Author(s):  
Jesmeen Mohd Zebaral Hoque ◽  
Jakir Hossen ◽  
Shohel Sayeed ◽  
Chy. Mohammed Tawsif K. ◽  
Jaya Ganesan ◽  
...  
Keyword(s):  
Incomplete Data ◽  
Missing Values ◽  
Prediction Models ◽  
Low Cost ◽  
Support Vector ◽  
Data Sampling ◽  
Data Set ◽  
Missing Value ◽  
Missing Value Imputation ◽  
Proper Analysis

Recently, the industry of healthcare started generating a large volume of datasets. If hospitals can employ the data, they could easily predict the outcomes and provide better treatments at early stages with low cost. Here, data analytics (DA) was used to make correct decisions through proper analysis and prediction. However, inappropriate data may lead to flawed analysis and thus yield unacceptable conclusions. Hence, transforming the improper data from the entire data set into useful data is essential. Machine learning (ML) technique was used to overcome the issues due to incomplete data. A new architecture, automatic missing value imputation (AMVI) was developed to predict missing values in the dataset, including data sampling and feature selection. Four prediction models (i.e., logistic regression, support vector machine (SVM), AdaBoost, and random forest algorithms) were selected from the well-known classification. The complete AMVI architecture performance was evaluated using a structured data set obtained from the UCI repository. Accuracy of around 90% was achieved. It was also confirmed from cross-validation that the trained ML model is suitable and not over-fitted. This trained model is developed based on the dataset, which is not dependent on a specific environment. It will train and obtain the outperformed model depending on the data available.

scholarly journals MODIFIED POSSIBILISTIC FUZZY C-MEANS ALGORITHM FOR CLUSTERING INCOMPLETE DATA SETS

2021 ◽  
Vol 61 (2) ◽  
pp. 364-377
Author(s):  
. Rustam ◽  
Koredianto Usman ◽  
Mudyawati Kamaruddin ◽  
Dina Chamidah ◽  
. Nopendri ◽  
...  
Keyword(s):  
Experimental Data ◽  
Incomplete Data ◽  
Missing Values ◽  
Complete Data ◽  
Noise Sensitivity ◽  
Data Sets ◽  
Fuzzy C Means ◽  
Number Of Iterations ◽  
Fuzzy C Means Algorithm

A possibilistic fuzzy c-means (PFCM) algorithm is a reliable algorithm proposed to deal with the weaknesses associated with handling noise sensitivity and coincidence clusters in fuzzy c-means (FCM) and possibilistic c-means (PCM). However, the PFCM algorithm is only applicable to complete data sets. Therefore, this research modified the PFCM for clustering incomplete data sets to OCSPFCM and NPSPFCM with the performance evaluated based on three aspects, 1) accuracy percentage, 2) the number of iterations, and 3) centroid errors. The results showed that the NPSPFCM outperforms the OCSPFCM with missing values ranging from 5% − 30% for all experimental data sets. Furthermore, both algorithms provide average accuracies between 97.75%−78.98% and 98.86%−92.49%, respectively.

Lsquare System for Mining Logic Data

2011 ◽  
pp. 693-697 ◽  
Author(s):  
Giovanni Felici ◽  
Klaus Truemper
Keyword(s):  
Machine Learning ◽  
Data Mining ◽  
Pattern Recognition ◽  
Medical Diagnosis ◽  
Missing Values ◽  
The Other ◽  
Training Data ◽  
Patient Records ◽  
Mining Community ◽  
The Many

The method described in this chapter is designed for data mining and learning on logic data. This type of data is composed of records that can be described by the presence or absence of a finite number of properties. Formally, such records can be described by variables that may assume only the values true or false, usually referred to as logic (or Boolean) variables. In real applications, it may also happen that the presence or absence of some property cannot be verified for some record; in such a case we consider that variable to be unknown (the capability to treat formally data with missing values is a feature of logic-based methods). For example, to describe patient records in medical diagnosis applications, one may use the logic variables healthy, old, has_high_temperature, among many others. A very common data mining task is to find, based on training data, the rules that separate two subsets of the available records, or explains the belonging of the data to one subset or the other. For example, one may desire to find a rule that, based one the many variables observed in patient records, is able to distinguish healthy patients from sick ones. Such a rule, if sufficiently precise, may then be used to classify new data and/or to gain information from the available data. This task is often referred to as machine learning or pattern recognition and accounts for a significant portion of the research conducted in the data mining community. When the data considered is in logic form or can be transformed into it by some reasonable process, it is of great interest to determine explanatory rules in the form of the combination of logic variables, or logic formulas. In the example above, a rule derived from data could be:if (has_high_temperature is true) and (running_nose is true) then (the patient is not healthy).

Sign in / Sign up

Export Citation Format

Share Document