Review On a Bachelor’s Thesis of Irina Grigorieva “Canonical analysis of categorical data with application in marketing” The thesis of Irina Grigorieva is dedicated to the problem of establishing a relation between two sets of nominal variables. This is a common problem in the wide range of disciplines including biology, medicine, economics and social sciences. The thesis concerns a specific case of establishing the relation between characteristics of innovative projects and their future survival and success. Uncertainty coefficient is used as a measure of such relation. In addition, to identifying the strongest interdependencies between two subsets of variables it is suggested to use symptom analysis in order to determine which of them are the most stable. The work is not without some weaknesses. It is occasionally written in a colloquial rather than academic style. The literature review is thin if not absent. There are no examples of alternative approaches that were applied in a similar context. The extreme distribution of table 2.11 probably results from some data error and should be fixed or commented on. It could be also recommended to account for multiple hypothesis testing using Bonferroni or Šidák correction. Future development of the work may include non­parametric estimation of statistical significance of proposed procedure. One example is a permutation test. By reshuffling values of nominal variables and repeating proposed procedure on a permuted data set sufficient amount of times it is possible to compare the obtained results with the distribution of the results under the null hypothesis. During the preparation of the thesis, Irina Grigorieva did a large amount of high-quality work. She demonstrated knowledge of a wide set of statistical methods including both classical and recently developed. She also demonstrated her ability to use and combine them appropriately based on requirements of a particular problem. She performed an exploratory analysis that helped to evaluate the quality of data and provide insights into potential inconsistencies. The methodology was described precisely and in full details. A special software was developed to perform all necessary computations.The results clearly described and interpreted and thus provide the answer on the initial question. Taking all of the above into account I recommend the grade 4 out of 5. Ivan Smirnov, MSc Doctoral researcher National Research University Higher School of Economics