“Fake news” or “fake article” refers to a made-up narrative intended to mislead readers for individual or organizational gain in this era of digital journalism. The truth regarding current events and occurring has become seriously muddled due to fake news. Fake news is one of the main problems the journalism sector is facing. Identifying if a story or article is real or a fraud is an automated method known as fake news detection. By using a set of classifiers at two levels, we presented in this work an effective level-based ensemble approach for identifying false news. Based on K4 cross-validation (75% training, 25% testing), our experiment results show that the suggested ensembled method using the tf-idf feature extraction strategy achieved improved classification accuracy results. We made use of a Kaggle cohort that is free and open-source. There is a fictitious class with 17903 unique samples and a genuine class with 20826 unique samples in it. AUCs of 0.95 and 99.03% were obtained by Model 0 (Model 1) - KNN, Support vector, Decision tree, level-1 Logistic regression, and 94.43% and 4.64%, respectively, by Model 2 (level 0) - Support vector, level-1 Logistic regression, which outperformed Model 1 by 4.64%. The bogus news class was identified in less than 5 s using our suggested methodology.