% FROM: http://pipelines.local:9001/p/data_tree_biology
% SCION: failure


In contrast to pruning, a number of recent studies have suggested that predictive accuracy may also be improved by more complex partitioning of an instance space than that formed by standard decision tree induction, using the grafting technique. Once the tree is constructed, then the new branches that can be added to the tree, which are grafted to improve the decision making process. Pruning and Grafting are complementary methods to improve the decision tree in supporting the decision. Pruning allows cutting parts of decision trees to give more clarity and grafting adds nodes to the decision trees to increase the predictive accuracy. To achieve grafting new branches can be added in the place of a single leaf or graft within leaves. However, the potential new branches have to be selected carefully to avoid increasing the error and failure cases.
Grafting in biological trees is also a useful technique. The method consist in asexual plant propagation that join plant parts so they will grow as one only structure. Some varieties of plants do not come true from seeds. The seed from a Haralson apple will produce an apple tree, but it will not produce a Haralson apple tree. Fruit trees cannot be reproduced "true" to the original cultivar from seed, they can only be reproduced by grafting. This process uses a root system better adapted to soil or climate than that produced naturally by an ungrafted plant
Although grafting can be successful if the two parts of the plants are compatible (scion on top and rootstock on the bottom), there are many reasons why a graft can result in failure:
	* Scion and rootstock not compatible (not fitting pipes) 
	* Not meeting properly of the vascular system of each plant (leakage) 
	* Scion is upside-down (contrary flow direction)
	* Grafting is performed at the wrong season (influence of weather and time)
	* Scion and rootstock are unhealthy
	* Scion is dried or injured by cold 
	* The union is displaced by storm, birds, or other means (leakage) 
% [_data_tree_biology_](http://pipelines.local:9001/p/data_tree_biology)