Imputation techniques for the reconstruction of missing interconnected data from higher Educational Institutions

Abstract

Educational Institutions data constitute the basis for several important analyses on the educational systems; however they often contain not negligible shares of missing values, for several reasons. We consider in this work the relevant case of the European Tertiary Education Register (ETER), describing the Educational Institutions of Europe. The presence of missing values prevents the full exploitation of this database, since several types of analyses that could be performed are currently impracticable. The imputation of artificial data, reconstructed with the aim of being statistically equivalent to the (unknown) missing data, would allow to overcome these problems. A main complication in the imputation of this type of data is given by the correlations that exist among all the variables. We propose several imputation techniques designed to deal with the different types of missing values appearing in these interconnected data. We use these techniques to impute the database. Moreover, we evaluate the accuracy of the proposed approach by artificially introducing missing data, by imputing them, and by comparing imputed and original values. Results show that the information reconstruction does not introduce statistically significant changes in the data and that the imputed values are close enough to the original values.

Optimization methods for the imputation of missing values in Educational Institutions Data

“…The proposed methodology appears capable to reconstruct the information of the missing values without introducing statistically significant changes in the dataset, and the imputed values result to be close enough to the original values, as shown by experiments in [3] .…”

“…It imputes the Recipient Institution by using the values taken from a similar institution, called Donor, selected using optimization criteria. The full mathematical details of the methodology are explained in [3] . The original ETER dataset, possibly integrated with bibliometric information, and the imputed dataset, are available in several variants from [5] .…”

“…This work describes in detail the methods and the codes implementing the methodology proposed in [3] to impute missing values in interconnected data. In particular, a first part of the proposed methodology, called ``trend smoothing imputation'', is designed to impute missing values in time series by respecting the trend and the other features of an Institution.…”

Determinants of the incidence of non-academic staff in European and US HEIs

“…12 The imputation method is based on Bruni et al (2021) from which the imputed data were taken. 13 Previous studies included in the relevant literature consider the number of non-academics relative to the number of academics/total staff (Baltaru and Soysal 2018;Gornitzka and Larsen 2004).…”

Denoising autoencoder-based imputation of Harvard-MITx person-course dataset

“…The above is the initial exploration of experts and scholars on missing value processing in the field of education, but the missing value processing method used does not consider the correlation between data. A recent study [32] used techniques such as weighted mean, linear regression, trend smoothing, etc. to deal with missing values in European higher education datasets, and while achieving good performance, ignored local correlations between the data.…”