Description and Application of Linear and Non-Linear in Data Mining

Data mining or data mining is a method for finding patterns, features, and relationships between data automatically. Data mining can be applied in various fields such as business, marketing, science, even the world of art and entertainment. Based on its functionality, data mining can be grouped into several groups, one of which is classification.

Classification method can literally be interpreted as a process that aims to find a function or model to distinguish a data class from objects whose labels are not yet known. The classification process starts from learning a data / training process that aims to build a model based on certain rules or functions. The classification itself has many techniques in finding labels from a class, including the Bayes Classification, Decision Tree, Artificial Neural Network, Support Vector Machine, Nearest Neighbor Rule, and Fuzzy Logic Based Classification.

In general, there are many problem areas that require a classification method to solve, such as in customer management, predicting sales volume, medical analysis, marketing, including fraud detection and so on.

There are two methods to solve a problem, namely linear and nonlinear. Linear programming is the application of mathematical methods in solving problems in the allocation of limited resources which aims to maximize profits and minimize production costs.

Meanwhile, nonlinear programming is another alternative if a problem cannot be solved by linear programming. Nonlinear functions can generally be found in the multiplication of 2 or more independent variables, n-power functions with n> = 2, exponential functions, trigonometric functions and logarithmic functions.

In short, linear and nonlinear algorithmic modeling is an operations research technique that aims to solve an activity plan that has been formed in a mathematical model in order to solve a problem.

Examples of applying linear and nonlinear programming are as follows. In any company, in any field, it certainly has limitations, both in terms of the number of raw materials, equipment that supports the running of a business, human resources, working hours, including how much capital it must spend. All of these problems need to be taken into account so as not to make the company lose money through optimization by maximizing profits or minimizing costs.

In one case, there are things that cannot be solved by the linear method because certain factors are the cause of the nonlinearity in the objective function. For example, if the company is facing price elasticity, there are many goods that are sold that are not directly proportional to the price or it can be said that the more limited the product, the more expensive the price is. In this case, the company cannot solve it linearly and must use the nonlinear method because there are other factors that need to be taken into account.

Another example of modeling linear and nonlinear functions used in data mining to solve a problem in classification, one of which is the Support Vector Machine. The SVM system used in learning is based on optimization theory. SVM itself is a flagship in the field of pattern recognition. Based on its characteristics, the SVM method is divided into 2, namely linear and nonlinear. Linear modeling separates the two classes on hyperplane and soft margin. It is different from nonlinear, which is a function of the kernel trick on high dimensional spaces. The application of SVM classification uses linear and nonlinear functions such as Public Analysis Sentiment on the Jakarta Government Lockdown Policy Using the SVM Algorithm, Spam Classification on YouTube Using the Naïve Bayes Method, Support Vector Machine, and K-nearest Neighbors, Support Vector Machine Methods and Forward Selection Payment Predictions Purchase of Copra Raw Materials.