Know more

After having brought together and selected the raw data, a work of translation of these data had to be undertaken to match the expectations of the model, in terms of qualitative scales for the basic attributes. Indeed, the data provided may be of a very different nature and it must be taken into account in this diversity. For example, for the winter wheat, a sowing rate may be registered in number of grains per hectare,or in kilogram of seed per hectare or in ordinal value : high, normal, low. The chosen scales for some basic attributes are synthesized on the attached table. Some scales use the term "favourable", it must then be understood as favourable to the pest, and vice versa for "unfavourable".
Thus, either the data collected are qualitative, ordinal or nominal. In this case, the quantitative values have already been assessed qualitatively by the experimenters in function of the regional context and it is recognized then that their expertise are admissible.
Either the data collected are quantitative and in this case, we have applied rules of decision, defined for the regions represented in our data set and based on the technical literature and expertise. These decision rules are grouped in an "interface", not directly visible to the user but quite transparent.
These data were subsequently used to assess the predictive quality of sub-models by comparing the simulated intensities with the observed intensities. Several  statistics indicators have been used as evaluation criteria.