Amount of information in the form of online news needs to be balanced with the ability of readers to sort or classify subjective or objective news. So that a special system is needed that can be used for online news objectivity classification so that it can help readers to pick up subjective or objective news. This research proposes the development of techniques in machine learning to help sort out news objectivity automatically based on the content of the news. The algorithm proposed is K-Nearest Neighbor (KNN) algorithm. News samples obtained from kompas.com by scrapping occur imbalance classes where the number of objective news and subjective news are not balanced. So that it can affect the performance of the classification algorithm. One technique to overcome the imbalance class is to apply the Synthetic Minority Over-sampling Technique (SMOTE) technique.. SMOTE is the generation of minority data as much as the majority data. This study compares the performance of  KNN algorithm without SMOTE and the performance of KNN algorithm with SMOTE. Based on the results of the study by applying a variety of neighboring k values, namely 1, 3, 5, 7 and 9, it was found that the application of SMOTE could improve the accuracy of the KNN algorithm at values ​​k = 1 and k = 3 with an average increase of 3.36. At values ​​k 5, 7 and 9 the algorithm experiences an average decrease in accuracy of 6.67.

People now trying to maximizing function of Virtual Learning Environment. Virtual Learning Environment, not only as a place to help learning system but now has become a place of learning itself. But with the change of the learning system, teacher now have difficulty to monitor the activity of the student and the learning material. Although there is data that is considered capable become a benchmark for students and the interaction with Virtual Learning Activity. This paper will make a data prediction using Naïve bayes and C4.5 Algorithm using the Web History data and the sum of webpage interaction of the students in Virtual Learning Environment.

CNN originates from image processing and is not commonly known as a forecasting technique in time-series analysis which depends on the quality of input data. One of the methods to improve the quality is by smoothing the data. This study introduces a novel hybrid exponential smoothing using CNN called Smoothed-CNN (S-CNN). The method of combining tactics outperforms the majority of individual solutions in forecasting. The S-CNN was compared with the original CNN method and other forecasting methods such as Multilayer Perceptron (MLP) and Long Short-Term Memory (LSTM). The dataset is a year time-series of daily website visitors. Since there are no special rules for using the number of hidden layers, the Lucas number was used. The results show that S-CNN is better than MLP and LSTM, with the best MSE of 0.012147693 using 76 hidden layers at 80%:20% data composition.

No abstract

In Kediri City there is a very popular woven fabric shop called Medali Mas. It has high sales transaction activity resulting in a large stack of data purchases. This data stack is examined as an information pattern for consumer purchases using data mining association rule techniques and FP-Growth algorithms. The FP-Growth algorithm uses the concept of development tree in searching for frequent item sets. The data used are, 26 types of woven fabric items and 200 transaction data provided that 2 or 3 types of items in 1 transaction. Determined  minimum support value of 20 percent and  minimum confidence value of 10 percent. It also used Chi-Square testing to find out how much correlation between variables from the results of frequent itemsets that have been calculated. The final result of the consumer purchasing pattern is obtained (m to no) when buying Semi sutra Lusi = grey, Pakan = Blue Flowers, then the consumer might buy Sarong Lusi = black, Pakan= green Lurik and Cotton Lusi= yellow, Pakan = Tosca Bamboo with the results of the correlation between variables at 19.1397274913.