Vliv mediální komunikace a sezónních faktorů na návštěvnost poboček firmy a na návštěvnost webu firmy

We invesigate the impact of online and offline advertising on footfalls and web visits of unspecified company. The influence of web visits on footfalls is also measured. Model explaining footfalls is build in marketing QED program. Positive influence in the model had TV communication, web visits, 4 different sales and November. Negative impact had TV communication of 3 competitors, temperature, public holiday of st. Wenceslas, order of week in a month and February. This model was also estimated in R where we used nonlinear regression and GAM. For estimating the initial level of parameters in nonlinear regression two approaches were used. One with auxiliary regressions for each variable and second by builing bounds based on marketing QED results. Using genetic algorithms we tunned GAM and nonlinear regression parameters to minimize deviance of models. We also tunned GAM by minimizing BIC. None of these approaches signifficantly improved analyzed models, orginal GAM without tunning proved to be the best from models estimated in R. All models on footfalls were compared by AIC, BIC and MAPE, when model from makreting QED proved to be the best. Web visits were also modeled in marketing QED. Positove impact had variables communication in TV, radio, press, online ads on Facebook, Seznam and Google, online blog visits and 6 differend sales. Negative impact had temperature, 1 public holiday and week order. In R we estimated GAM and tuned it by GA, which led to better results. In model comparison AIC and BIC was better for model from marketing QED but MAPE was better for GAM.