Xxii республикалық студенттер мен жас ғалымдардың ғылыми конференция материалдары
жүктеу/скачать 8,13 Mb. Pdf көрінісі
|
Сборник материалов конференции (продолжение)
- Бұл бет үшін навигация:
- Decision Tree
| K-Nearest Neighboars (KNN)
K-Nearest Neighboars is very simple for understanding method, the goal of knn in medicine is usually classification of classes, as in logistic regression. The object is assigned to the class that is the most common among k neighbors of this element, the classes of which are already known. On the stage of learning the algorithm remembers all the feature vectors and their referenced defined classes. After that algorithm tries to deal with unknown observations from dataset. Then, k vectors closest to it are selected, and the new object belongs to the class to which most of them belong (Figure3). Practical example of using this method is classification of lymph node metastasis in gastric cancer[3]. 229 Figure 3 Decision Tree Decision tree is a solution to the problem of learning with a teacher, based on how people solve forecasting problems. In the general case, this is a tree with decisive rules in non-leaf vertices (nodes) and some conclusion about the objective function in leaf vertices (forecast). The decisive rule is a certain function of the object, which allows you to determine which of the child vertices you want to place the object in question. Different objects can be in leaf vertices: the class that needs to be assigned to the object that got there (in the classification problem), class probabilities (in the classification problem), and directly the value of the objective function (regression task). Most often, binary decision trees are used in practice (Figure 4). Figure 4 In medicine, this technique helps to make the right choice of options for action when solving clinical problems, especially if available options come with high costs and risks. Constructed decision tree performed by the graphical method compares alternative options and the results of medical interventions obtained by analyzing costs, risks and utility[4]. жүктеу/скачать 8,13 Mb. Достарыңызбен бөлісу:
|