We consider the accuracy estimation of a classifier constructed on a given training sample. The naive resubstitution estimate is known to have a downward bias problem. The traditional approach to tackling this bias problem is cross-validation. The bootstrap is another way to bring down the high...

One hundred years ago Karl Pearson derived an approximate formula for the correlation between ratios with a common divisor and cautioned to be wary of correlating ratios. The exact formula for the correlation between ratios is derived. It can provide a better reference point of no connection...

The proportional odds model is a popular regression model for ordinal categorical responses, which has a rather strong underlying assumption, the proportional odds assumption. The rejection of the null assumption, however, is not very informative since a statistical significance does not...

Despite its long history in antitrust policy, predation remains a poorly understood phenomenon. The main difficulty is in empirically identifying predatory intent. I propose a method for measuring the effect of reputation, whose significance would enable us to infer predatory intents. Maximum...

It is well known that patent pools can enhance efficiency by eliminating the complements problem. This paper investigates how the presence of vertically integrated firms affects the economic impact of a patent pool. Without a patent pool, the presence of integrated firms may either increase or...