The article deals with a special case of the preparation of data about the vehicles movements which comes in large volumes from the source to the accelerated applied methods of data mining. Data preparation goes through several stages from selecting the necessary fields and records to saving them with modified values into a new data structure. The source data which consist of 18 fields has a share of incorrect information and formats of numerical information that are not suitable for further processing. The source data is large in volume and processing it in the original form takes a very long time. The article shows how to use the pthreads library to organize multi-threaded processing of this data. To confirm the applicability of this library, the article presents the results of numerical experiments.