This Thesis is dedicated to the research of the problem of processing short time-series and their characteristic parameter data by applying data mining methods and algorithms to determine the prospective future value of the analyzed object. The goal of the Thesis is to develop a data mining based set of approaches for the processing of short time-series and their characteristic parameters in a forecasting system that would be applicable in various fields and would only use the characteristic data of a analyzed object in order to determine the forecast. The process of analysis of short time-series and their parameters is hard to formalize because the process of finding patterns in these data is very complex. The algorithms and approaches applied in such task would have to perform well with different data structures. The process of short time-series analysis is formalized in the Thesis as a clustering task; whereas the process of descriptive parameter analysis is formalized as classification task. Processing of short time-series is carried out using an adapted k-means algorithm, which allows determining the most eligible number of clusters based on the mean absolute clustering error. New approaches are proposed for merging of clustering results and characteristic parameters. The developed forecasting systems also include implementations of new approaches to the analysis of clustering and classification results. The Thesis offers three different systems for short time-series and their characteristic parameter processing that are adapted to different fields of application: sales, pharmacology and healthcare. The first system makes forecasts of product sales volumes for a following period of time based on the characteristic parameters of a new product entered into the system. The second system is used to forecast the risk of heart necrosis based on the characteristic parameters of a analyzed individual. The third one carries out the forecasting of bacterial proliferation syndrome in the small intestine based on the characteristic parameters of health self-assessment of a respondent. The performance and functioning of the system is tested in practical fields with real data. The use of algorithms and approaches employed in the system implementation is experimentally substantiated. The analysis of the system and approach collection developed and proposed in the Thesis for different fields offers guidelines for development of similar systems that can be used to process data of short time-series and their characteristic parameters. The thesis is written in Latvian. It contains an introduction, 4 chapters, conclusions, the list of references, 2 appendixes, 53 pictures, 25 tables, 144 pages. The list of references contains 77 records.