| Abstract [eng] |
In order to warrant the quality of statistical information the Department of Statistics to the Government of the Republic of Lithuania follows Quality declaration of the European Statistical System and Code of practice for European statistics. The Division of statistics on industry, accordingly, in order to improve the quality of information it publishes may increase the number of primary statistical data controls and error eliminations. Thus, the primary aim of this thesis is to improve the quality of data (on the amount and changes of production and trade of major products) that is published by the Division of statistics on industry, i.e. to automate the search for the logical errors in the reports on the production of industry products at enterprises. A software application was developed for the elimination of logical errors. It was tested with the set of primary statistical data (the data before the secondary editing) and it provided as its result a list of companies that have made some kind of mistake. By means of logistic regression it was found that the appearance of a mistake is conditioned by the method of submission of the report (i.e., the uploading of an electronic form), by the level of aggregation of a product code (i.e., the codes that resolve into smaller codes) and by the measures when they are “-”. |
