Leveraging Hadoop and Hybrid Deep Learning on Home Datasets for Business Intelligence

In today's data-driven business environment, organizations' dependence on cutting-edge technologies is constantly increasing in order to gain meaningful information from big data. Raw data's size and complexity make it unusable for decision-making. To address this, BI systems transform raw data into clear, insightful information by leveraging the Hadoop framework to process big data. This study is concentrated on the creation of a business intelligence (BI) system depending upon deep learning (DL) approaches, especially one-dimensional convolutional neural networks (1-D CNN) and long short-term memory (LSTM). The proposed approach takes advantage of the DL algorithms for examining and picking up intricate patterns in sequential data, which is helpful in accurately anticipating the results and offering relevant insights. The predictive capabilities of the proposed system are enhanced through a combination of the 1D CNN and LSTM models, enabling it to grasp spatial and temporal data dependencies. Parallel computing and distributed processing made possible by the Hadoop model have increased the efficiency of big data management, which ensures performance and scalability while working with large datasets. This study aims to show how well the proposed business intelligence system is based on hybrid deep learning to make predictions by utilizing big data analyses. The results show the superiority of the integrated CNN-LSTM model, which operates on a data block size of 512 MB, over a data block size of 64 MB for home data sets, in addition to its superiority over two machine learning models (decision tree and booster regression) at the same block size.