Applying Machine Learning Techniques for Environmental Reporting

Environmental Accounting progress in Greece is relatively slow in comparison to the more developed countries as only recently the national legislation system adopted ´environmental friendly´ standards. This paper seeks to identify, for the first time, the level at which Greek listed companies from several sectors provide environmental information through their financial statements. Moreover, we intent to discover the level at which environmental reporting is determined by the information position as explained by information cost variables, proprietary cost, control variables and media visibility variable. For this reason, we compared a number of different machine learning models and came to the conclusion that an ensemble of models gave more accurate results.