Methodology

The HCSO takes over on a quarterly, season-related basis from the Hungarian Public Roads PLC data from their cycle traffic measuring points within the OSAP (National Data Collection Programme) data receiving no. 2591 framework. Our study included measuring points from around the great lakes of Hungary (Lake Fertő, Lake Tisza, Lake Velence, and Lake Balaton).

Measuring devices count the number of passing by cycles. For making things simpler, one passing by was counted as one cyclist. In practice differences may occur, e.g. one adult with a child on a seat or tandem cycles.

We analysed the measuring points’ data of Fertőd (Lake Fertő), Tiszafüred (Lake Tisza), Balatonederics, Zánka, Örvényes, Keszthely (Lake Balaton), as well as of Pákozd, Agárd (Lake Velence) aggregated, on time series, monthly frequency basis and compared these with the statistical data of tourist accommodation establishments. Raw cycle traffic data may be used in a limited way, as only a few measuring devices worked reliably for a longer period of time. Their repair, expansion continues, making more accurate information available regarding a longer period of time.

For obtaining the best possible data supplement, machine learning (e.g. missForest) and time series models have been examined. Based on the results the ARIMAX model, using accommodation establishments data as auxiliary variable proved to be the most accurate; in consequence we supplemented missing data on a monthly basis by using this model. Due to the experimental character of the data supplement method we publish, for the time being, results quarterly, on a season-related basis.