Growing stock is an ecological indicator of forest ecosystem response to natural and anthropogenic impacts that may result from forest management measures or environmental impacts. Information on growing stock is thus essential to understand dynamics of forest stands, their productive capacity and to manage their use within limits of sustainability. Dynamic changes of forest growing stock, as well as predictions of their future development, are usually estimated from the data gathered by national forest inventories using some mechanistic modelling approach. The resulting models are informative, but include many parameters, some of which are difficult to set or estimate. Due to the demanding parameterisation of mechanistic models, it is hard to achieve stability of their output accuracy, which can lower their predictive power. This paper presents an alternative and complementary approach of constructing models with machine learning and data mining methods. We applied these methods to the Silva-SI database and used the resulting interpretable models in order to find explanations for structural changes in Slovenian forests over the period from year 1970 to 2010. In addition, we developed predictive models for growing sock in the decade from year 2010 to 2020. The structure of the models describing temporal dynamics of growing stock shows that trends of growing stock are increasing for the entire studied period, while accumulation of growing stock is much more intensive after 1990. Forests with a lower growing stock are located either in the areas with non-favorable site conditions for forest growth, or at lower altitudes, where they are more exposed to human exploitation due to their vicinity to more densely populated regions. Predictions of growing stock for the decade 2010–2020 suggest that Slovenian forests will continue to accumulate their growing stock (private owned forests to 327m3/ha and state owned forests to 343 m3/ha in 2020). The presented data mining approach that was here applied to the growing stock can also be used for investigating other ecological indicators.