ANALYSIS OF DATA COLLECTION PROBLEMS IN INTEGRATED STRUCTURES AND THEIR IMPACT ON THE ACCURACY OF PREDICTIVE ANALYTICS IN BUILDING BUSINESS PROCESSES

The effectiveness of business processes in the modern world is determined by the quality of big data analytics. These processes represent a sequence of interrelated actions aimed at achieving strategic and operational goals. One of the key obstacles to high-accuracy analytics is the heterogeneity of data sources, incompatibility of formats, delays in updating information, as well as errors in aggregation and processing. These factors provoke data distortion, which negatively affects the reliability of predictive models and can lead to inefficient management. BigData tools allow to conduct detailed business operations analysis, identify bottlenecks, optimize processes, and determine promising development directions. The efficiency of such solutions directly depends on the quality of the source data. In case of errors, inconsistencies or contradictions, predictive models lose their accuracy, which reduces the reliability of analytical conclusions. To minimize the risks associated with information distortion, it is necessary to standardize data collection, verification and processing procedures. The use of modern machine learning algorithms and statistical analysis methods allows to automatically detect errors, clean data and increase forecast accuracy. By optimizing data collection and analysis processes in integrated structures, companies can make informed decisions based on up-to-date information. This helps to increase competitiveness, create in-demand products, formulate effective strategies and quickly adapt to market changes. The impact of data collection problems in integrated structures on the accuracy of predictive analytics is considered. The key factors leading to data distortion are analyzed, including heterogeneity, incompleteness, outdatedness and integration problems. The need to standardize data collection, cleaning and verification processes to increase forecast reliability is substantiated. Modern machine learning and statistical analysis methods, the efficiency of which depends on data quality, are outlined.