<![CDATA[The study aims to determine the impact of the Logistic Regression method on the classification of customer bank deposits, using a public UCI Bank Marketing dataset, which contains customer-specific information of bank deposit telemarketing activities. Data has a binomial label consisting of 'yes' for subscribers and 'no' for non-subscribers. The data preprocessing phase is done with downsampling to make the amount of data more symmetrical, then data selection and data transformation to ensure that the data used values are consistent, attribute selection to select the attributes most accurately used and give significant influence. Classification is done using the Logistic Regression algorithm. Data is shared using a split method with 90% training data and 10% testing data, with the aim of optimizing the training process. The performance result consists of an accuracy of 88.53%, a classification error value of 11.4%, can be categorized as low, showing only a few errors produced by the algorithm model, a kappa value of 0.68 close to 1, so it is categorized well, a low RMSE rating of 0.3 indicates a model accurate, and a high AUC percentage of 93.4% indicates the correct algority used in this dataset, because it produces a good performance value.]]>
Copyrights © 2024
