Search

Explorative data analysis – interactive learning



Explorative data analysis – interactive learning

  • Explorative data analysis is increasingly applied in all scientific areas. The analysis of empirical data represents a key qualification which has to be tediously acquired by mathematicians, natural scientists, engineers, social scientists and economic scientists, either during their studies or in the course of initial research projects.
  • Statistical questions are often considered to be confusing and difficult. For many students data analysis constitutes an abstract process which underlies firm regulations without any constructive or explorative elements. Statistical methods are applied in a purely mechanical manner without any intuitive approach. This leads to the fact that data analyses do not represent any creative constructive acts which result in interpretable and significant statistical models.
  • Modern mathematical approaches to data analysis, for example the interactive statistical graphics, are hardly applied. Thus, many scientists are not equipped with the central tools necessary to judge the quality of the data. Explorative data analysis – backed by logical considerations – enables scientists to find unknown connections in the data and problems during the survey or with respect to data quality. Due to a mechanic approach in the data analysis, gross misinterpretations often occur, since the user is not aware of the rules of translation from reality into the data set, ie into the language of mathematics. The interpretation of a statistical analysis is nothing but a retranslation from the language of mathematics into reality. Such a translation, however, will only be correct if the rules of translation are clear and known to the user from the very beginning.
  • It is of major importance for students of explorative data analysis to learn how to use the tools of data analysis, handle real data sets and interpret them in their real environment. Didactic data sets lacking a connection to a real problem are not suited to represent the principles of explorative data analysis and to ensure that students really understand this method. In order to process real data sets it is indispensable to apply interactive statistical graphics to real data sets. In this context, interactivity does not only mean menu-controlled user guidance but a direct dependence (dynamic connection) between the individual application windows. That is to say changes made in one window directly effects the other windows. This interactivity can only insufficiently be demonstrated and explained by manuals, tutorials, help functions of programmes or static whole-class teaching.
  • Since the explorative data analysis is based on the computer and interactive software, it is suggested to use congruent E-learning concepts in teaching explorative data analysis. In interpreting difficult data sets it has become evident that it is important to cooperate with other mathematicians to allow a comprehensive analysis and interpretation of the data sets. For this reason, it is suggested to use cooperative didactical concepts for learning to use data analysis. In this respect, too, E-learning seems to be a suitable method, since students may exchange their ideas via forums, wikis, video conferences, etc and can work on a problem in a cooperative manner. Students are often confronted with the problem that their concepts of mathematics learned at school are unsuitable to form a skeleton for a new learning content. To re-adjust the learning skeleton it might be useful to use analogies. E-learning offers perspectives in this respect, too, since due to analogous structures elements of data analysis can be attributed to elements of an E-learning platform, which thus creates added value for the students.
  • The previously mentioned aspects led to the idea to provide an E-learning/Blended learning platform for students and experts (scientists) who want to learn explorative data analysis.

Scientists