The issues of work related to solving problems of intellectual data processing is to increase the efficiency of qualimetry, classification, diagnostics, choice, decision making, forecasting, taxonomy, etc., using data mining algorithms and a statistical approach. The aim of the work is to formalize the description and analysis of the source data as a means of improving the quality of solving intellectual problems. Compared with the well-known works, in which the emphasis was placed on certain informational criteria, the approach proposed in the work allows to form a set of quantitative and qualitative criteria for the formalization of information sources. To achieve the goal, taking into account the knowledge of the structure and context of the problem, objective requirements imposed on the input information have been formulated. To ensure the specified quality of solving the problems of intellectual data processing, it is necessary that the input information be objective, reliable, relevant, complete, timely, understandable, etc., and have a certain degree of accuracy, which is possible only with competent experts. The paper considers the concept of single and group expertise. The dependence of the quality of the solution of the data processing problem on the requirements for the source data is considered by the example of the forecasting problem. The accuracy of forecasting the results of external independent estimation in the selected section on the degree of completeness of the initial data and the selected methods of data processing and analysis is analyzed.