Innovation Indicator Analysis in the European Union: A Machine Learning Approach
| Author: | Bc. Jan Malecha |
|---|---|
| Year: | 2019 - summer |
| Leaders: | Petr Pleticha |
| Consultants: | |
| Work type: | Bachelors |
| Language: | English |
| Pages: | 60 |
| Awards and prizes: | |
| Link: | https://is.cuni.cz/webapps/zzp/detail/202930/ |
| Abstract: | The European Commission annually publishes a European Innovation Scoreboard (EIS) as a tool to measure the innovation performance of the EU Member States. This thesis extends the analysis published in the EIS 2018 in two different manners. The first part, a clustering analysis, examines the partition of the EU Member States to innovation performance groups. The thesis comes with a unique scheme of partition created by using hierarchical clustering. A comparison with the existing scheme shows that the general trends are similar in both schemes. The only main exception is the differentiation of the British Isles and Luxembourg apart from the other high performing countries. The proposed scheme provides insight about the within-cluster similarities, such as the similarity of Finland, Sweden and Denmark and their relative distinction from France, although they belong to one cluster. The second part, a regression analysis, attempts to examine the impact of innovations on real labour productivity. Contrary to existing literature, we do not find a statistically significant relationship between productivity and the components of the EIS. Additionally, the analysis is extended by the lasso estimation that provides a variable selection. The latter approach improves our findings and identifies four EIS indicators with positive impact on labour productivity |
