Introduction

A bioeconomy is key in reaching a sustainable economy. It promotes circularity and lowers dependence on non-renewable resources. Yet, there is great uncertainty on future availability of bioeconomic resources/biomass. 

The goal of this playbook is to provide a basic understanding of how classic econometric and novel machine learning modeling techniques influence forecasts of future availability of biobased materials.

Forecasting bioeconomy development has several challenges:

• Heterogeneity: Countries vary in agro-ecological conditions for growing biomass, presence of biomass processing industries, and policies (click on the text below for an illustration)
• Volatility:  Evolution of biomass acreages may be affected by sudden shocks, complicates prediction since it does not evolve in a linear way.
• Data sequentiality: biomass acreages evolve over time, prediction needs to take this time ordering into account
• Nonlinearities: Evolution of acreages over time may evolve in a non-linear way (e.g. leveling off, reversal). This could be due to the interactions of variables that change over time.
• Auditability: What are processes to control and audit the model.