Unsettled debt is a central global problem for banks and businesses. Being able to accurately predict which loans will default would be incredibly valuable for understanding the state of the economy. Machine learning models can be used to predict debt defaults, which can help these organizations make better decisions and reduce their risk. This study contributed a unique dataset collected from Egyptian Social Insurance organization (NOSI) with a proposed approach for defaulted credit prediction using classification learning and resampling methods for solving imbalanced data, which is a common challenge in debt default and credit data analysis studies. The approach was evaluated using three tree-based classification algorithms: Decision Tree (DT), Random Forest (RF), and eXtreme Gradient Boosting (XGBoost). The Similarity Oversampling and Under-sampling Pre-processing (SOUP) resampling technique outperformed all other resampling techniques when applied with RF, achieving high micro-average precision and recall of 66.67% and 67.38%, respectively, compared to the original dataset's 50.42% and 50.39%.

K-nearest-neighbor (KNN) is one of the state-of-the-art machine learning algorithms used for classification and regression tasks. In addition to being simple to understand, KNN is also versatile, spanning various applications. Despite its simplicity, it is considered a lazy classifier that does not generate a trained model but stores or memorizes training examples instead. Consequently, the prediction process using KNN becomes costly in resources and time, especially when the dataset becomes large. Also, there is no general way to choose the best distance metric during the prediction. This paper proposes a new algorithm called K-nearest Medoid KNN (KMKNN) which improves the performance of KNN in terms of prediction performance and time efficiency without a major effect on its result accuracy. The core idea of the proposed KMKNN is to cluster the dataset before the prediction to limit the distance measures to those data instances that belong to the nearest cluster of the new data. KMKNN when compared to the traditional KNN and other similar extended versions of KNN, achieves a noticeable improvement on 15 benchmark datasets. The importance of this work is primarily in large datasets or when the distance measure used is computationally expensive, which is common in the computer vision and pattern recognition domains.

Many machine translation studies have used large parallel groups to address sets of major European dialects. However, due to the lack of sufficient parallel information, few studies have considered Italian and Arabic. Moreover, dictionary-based translations of the Holy Quran from Arabic to Italian are usually incorrect. The meaning of the Quran has not been translated correctly. Because the dictionary-based translation considers the Quran to be a traditional document and translates it in order. This paper contributes in two ways. First, it presents a parallel corpus of 6237 Italian-Arabic sentences. Second, the paper introduces two deep learning models namely, long-shortterm memory (LSTM) sequence-to-sequence with an attention mechanism and Gated Recurrent Units (GRU) sequence-to-sequence with an attention mechanism for Arabic to Italian machine translation. Each of the proposed models is evaluated based on BLEU, ROUGE, and Cosine Similarity scores. The results indicate that the LSTM-based neural machine translation (NMT) outperforms the GRU-based NMT framework. The experimental results indicate that the LSTM model achieved mean scores of 0.96, 0.91, and 0.90 for Cosine Similarity, BLEU, and ROUGE, respectively. The GRU model achieved average scores of 0.94, 0.89, and 0.88 for Cosine Similarity, BLEU, and ROUGE scores, respectively.