Comments on Decision tree
At first glance, the decision making process can be seen as straightforward. This is because people make hundreds of decisions everyday, and often the same routine can be used for most decisions. For example, decisions concerning whether or not to brush your teeth: ‘If you wish to have clean teeth then you brush, otherwise don’t’. However, the decision of whether or not to brush your teeth is seemingly small and insignificant in comparison to the other decisions that people make everyday e.g. the purchasing of homes, changing occupations, migration overseas etc. These larger and lifestyle effecting decisions have hundreds of different factors and issues that need to be investigated and interpreted before a decision is made. In some cases, decisions can take up to days, even weeks.
Decisions Trees is a simplistic method of looking at decision making. All potential avenues have been identified and the advantages and disadvantages have been mentioned. These avenues are usually numerically mentioned and then calculated statistically to produce the most influenctial or probable cause of action. Although, decision trees and even its associated CART methodology, can not always infer real-world or external circumstances into account, thus forcing models to constantly change.
To be completely honest, after further researching artificial neural networks, I find them to be infinitely complex. Yes, the theory is simple enough. Individual processing systems that are interconnected with value of their weight (wi) determining their strength. However, Supervised, unsupervised and reinforced learning is utterly complex and the many different algorithms used for training neural network models… If anyone can simplify it for me, I would be quite thankful!
No comments:
Post a Comment