A classic supervised learning algorithm
KNN algorithm: determine which came from the sample and the sample class recently, then this sample belongs to this category.
Decision Tree: Which conditions on the result determined by the maximum entropy increment, it is determined that this condition is the first feature, and then determine which conditions the greatest impact on results by entropy increment on the basis of the first feature determines the second features, and so on, until entropy is zero, or no effect until the conditions.
Second, unsupervised learning algorithm
After specifying into K classes, randomly generated coordinate of the center K classes, using the k-th cluster center coordinates, the center coordinates of each class set after the cluster is calculated to update the center coordinates of the k, and then use the update: K-Means Algorithm k-th cluster center coordinates again, and then calculates the center coordinates of the cluster, and then update the center coordinates of the k. This is repeated until the center coordinates do not change or little change.
Apriori association rules algorithm: first calculate the probability of a series of frequent sets, and then calculate the support and confidence of the rule. (Mainly because I did not quite understand this algorithm can only learn a brief talk about)
