`coef` shows the coefficients for the rules and linear terms. In a binary classification problem, they are logistic regression coefficients. That is, the increase in log-odds of belonging to the target class, if the conditions of the rule apply. Rules with a negative coefficient decrease the probability of belonging to the target class if their conditions apply, rules with a positive coefficient increase the probability of belonging to the target class if their conditions apply. The larger the value of the coefficient, the larger the in- or decrease.