MOTIVATION In microarray gene expression studies, the number of replicated microarrays is usually small because of cost and sample availability, resulting in unreliable variance estimation and thus unreliable statistical hypothesis tests. The unreliable variance estimation is further complicated by the fact that the technology-specific variance is intrinsically intensity-dependent. RESULTS Th...

Recent research suggests that optimistically biased earnings forecasts issued by analysts are attributable to analysts minimizing symmetric, linear loss functions. We test an alternative explanation, namely that analysts have asymmetric loss functions. Theory predicts that if loss functions are asymmetric then forecast error bias depends on forecast error variance, but not necessarily on skewne...

We consider the problem of estimating a measure of daily volatility from intermittent high-frequency data that are subject to market microstructure effects. We show that a simple Newey-West type modification of the realized variance (RV) yields an unbiased measure of volatility for the ‘open’ part of the day. The modified RV is unbiased even if 1-minute intra-day returns are used. Further, with...

In a linear regression model with homoscedastic Normal noise, I consider James–Stein type shrinkage in the estimation of nuisance parameters associated with control variables. For at least three control variables and exogenous treatment, I show that the standard leastsquares estimator is dominated with respect to squared-error loss in the treatment effect even among unbiased estimators and even...

Total variance is a statistical tool developed for improved estimates of frequency stability at averaging times up to half the test duration. As a descriptive statistic, Total variance performs an exact decomposition of the sample variance of the frequency residuals into components associated with descending frequency octaves. As an estimator of Allan variance, Total variance has modest bias an...

Typical issues under consideration when selecting or designing a classification algorithm are the bias and variance components of error induced by the algorithm [1]. For example, one may choose a simple algorithm (e.g., linear combination of feature values, Naive Bayes, single logical rules, etc.) and draw a hypothesis from a small family of functions; the poor repertoire of functions may produ...

The ratio-type estimators have been introduced for estimating the mean and total population, but in recent years based on the ratio methods several estimators for population variance have been proposed. In this paper two families of estimators have been suggested and their approximation mean square error (MSE) have been developed. In addition, the efficiency of these variance estimators are com...

We study the off-policy evaluation problem— estimating the value of a target policy using data collected by another policy—under the contextual bandit model. We consider the general (agnostic) setting without access to a consistent model of rewards and establish a minimax lower bound on the mean squared error (MSE). The bound is matched up to constants by the inverse propensity scoring (IPS) an...