now fit the logistic regression model using a training data period from 1990 to 2008, with lag2 as the only predictor. compute the confusion matrix and the overall fraction of correct predictions for the held out data (that is, the data from 2009 and 2010).
