In the logistic regression model a set of 12 variables were used as input variables.
The Table 2 presents all of these variables with a description. Some of the
variables (OVOL8, ONO3, ONO6, and ONO9) had two representations in
different time steps. The first representative is from two time steps before the time
period at focus, the second is one time step before. The other four variables (age,
bank age, E_ind, and osuus_tr1) were collected from one time step before the
focused time period. In this case the volume and frequency variables could be
used as such because the analysis was carried out in different time periods
independently. The salary was used in the logarithmic scale because the impact it
has to the customers’ life is stronger among the customers earning small amounts
than among the customers that earn huge salaries.