August 2019
Intermediate to advanced
342 pages
9h 35m
English
In addition to the advantages already described, we must remember the possible disadvantages related to decision trees; these are essentially associated with the phenomenon of overfitting, which is due to the complexity of the tree data structures (it is in fact necessary to proceed in a systematic manner with the pruning of the tree, in order to reduce its overall complexity).
One of the undesirable consequences of the complexity is the high sensitivity of the algorithm to even the smallest changes in the training dataset, which can lead to sensible impacts on the prediction model. Therefore, decision trees are not the best fit for incremental learning.
Read now
Unlock full access