Search results

Machine learning methods have recently gained attention in business applications. We will explore the suitability of machine learning methods, particularly support vector regression (SVR) and radial basis function (RBF) approximation, in forecasting company sales. We compare the one-step-ahead forecast accuracy of these machine learning methods with traditional statistical forecasting techniques such as moving average (MA), exponential smoothing, and linear and quadratic trend regression on quarterly sales data of 43 Fortune 500 companies. Moreover, we implement an additive seasonal adjustment procedure on the quarterly sales data of 28 of the Fortune 500 companies whose time series exhibited seasonality, referred to as the seasonal group. Furthermore, we prove a mathematical property of this seasonal adjustment procedure that is useful in interpreting the resulting time series model. Our results show that the Gaussian form of a moving RBF model, with or without seasonal adjustment, is a promising method for forecasting company sales. In particular, the moving RBF-Gaussian model with seasonal adjustment yields generally better mean absolute percentage error (MAPE) values than the other methods on the sales data of 28 companies in the seasonal group. In addition, it is competitive with single exponential smoothing and better than the other methods on the sales data of the other 15 companies in the non-seasonal group.

Moving images represent a category of material which has historically received short shrift in most libraries and archives. Film, video, and now digital images form a significant part of many library and archival collections, however, and can be found in many formats and genres. Despite the ubiquity of such media in cultural institutions, the majority of libraries and archives owning collections of moving images have neglected these holdings—with the specific exception of those few archives devoted primarily to the care and preservation of moving images.

We document long-run trends in interstate migration rates, using individual-level data from the U.S. Census for the period 1850–1990. Two measures of migration are calculated. The first considers an individual to have moved if she is residing in a state different from her state of birth. The second considers a family to have moved if it is residing in a state different from the state of birth of one of its young children, allowing us to estimate the timing of moves more precisely. Overall migration propensities have followed a U-shaped trend since 1850, falling until around 1900 and then rising until around 1970. We examine variation in the propensity to make an interstate move by age, sex, race, nativity, region of origin, family structure, and education. Counterfactuals based on probit estimates of the propensity to migrate suggest that the rise in migration of families since 1900 could be explained by increased educational attainment, although education may be serving as a proxy for unmeasured covariates. The decline of interstate migration in the late nineteenth century remains to be explained.

Investors are inattentive to continuous information as opposed to discrete information, resulting in underreaction to continuous information. This paper aims to examine if the well-documented return predictability of the strategies based on the ratio of short-term to long-term moving averages can be enhanced by conditioning on information discreteness. Anchoring bias has been the popular explanation for the source of underreaction in the context of moving averages-based strategies. This paper proposes and studies another possible source based on investor inattention that can potentially result in superior performance of these strategies.

The paper uses portfolio sorting as well as Fama-MacBeth cross-sectional regressions. For examining the role of information discreteness in the return predictability of the moving average ratio, the sample stocks are double-sorted based on the moving average ratio and information discreteness measure. The returns to these portfolios are computed using standard approaches in the literature. The regression approach controls for various well-known return predictors.

This study finds that the equally-weighted monthly returns to the long-short moving average ratio quintile portfolios increase monotonically from 0.54% for the discrete information portfolio to 1.37% for the continuous information portfolio over the 3-month holding period. This study observes a similar pattern in risk-adjusted returns, value-weighted portfolios, non-January returns, large and small stocks, for alternative holding periods and the ratio of 50-day to 200-day moving average. The results are robust to control for well-known return predictors in cross-sectional regressions.

To the best of the authors’ knowledge, this is the first paper to document the significant role of investor inattention to continuous information in the return predictability of strategies based on the moving average ratios. There are many underreaction anomalies that have been reported in the literature, and the paper's results can be extended to those anomalies in subsequent research.

The findings of this paper have important practical implications. Strategies based on moving averages are an extremely popular component of a technical analyst's toolkit. Their profitability has been well-documented in the prior literature that attributes the performance to investors' anchoring bias. This paper offers a readily implementable approach to enhancing the performance of these strategies by conditioning on a straightforward measure of information discreteness. In doing so, this study extends the literature on the role of investor inattention to continuous information in anomaly profits.

While there is considerable literature on technical analysis, and especially on the performance of moving averages-based strategies, the novelty of this paper is the analysis of the role of information discreteness in strategy performance. Not only does the paper document robust evidence, but the findings suggest that the investor’s inattention to continuous information is a more dominant source of underreaction compared to anchoring. This is an important result, given that anchoring has so far been considered the source of return predictability in the literature.

Moves, or changes in query formulation, are made to resolve three problem situations: (1) when retrieved sets are too large; (2) when they are too small; or (3) when retrieved sets are off‐target. Observation and analysis of about ninety searches resulted in a list of eighteen operational moves, or modifications of query formulation, that keep the meaning of query components unchanged, and twelve conceptual moves which change the meaning of query components. All these moves are explained and then related to search tactics and strategies.

Activation is the process of preparing people, equipment and facilities for moving and start‐up. It is little understood, but integral to the success of facility. It starts well before and extends beyond the phases of commissioning and start‐up. Activation is described in six stages as it interacts with the traditional facility delivery cycle: long‐range planning; prioritization/budgeting; facility planning/design; facility completion; Operation; and criteria update/post‐occupancy evaluation. The outcome was a two‐volume, “user‐friendly” Activation Guide that describes activation for large complex projects in a straightforward way, and which is being used by the 172 major hospitals of the Department of Veterans Affairs in the USA. The process is generic and it could be adapted for use in activations of complex facilities in Europe.

The moving median method is suggested as an alternative for short‐term forecasting under some of the standard normal; Student t with degrees of freedom 1, 2 and 3; Cauchy; Chi‐square with degrees of freedom 1 and 2; exponential; and standard normal with outliers. When the Mean Absolute Deviation (MAD) and the Mean Square Error (MSE) are used as the criteria for evaluating forecasting accuracy, the moving average technique is superior to the moving median technique only for time series simulated from the standard normal distribution. The moving median technique is superior to the moving average technique for the fat‐tailed distributions; for example, t‐distributions of degrees 1, 2 and 3, Cauchy distribution, and the contaminated normal cases. An example shows the moving median technique responds to the level changes faster than the moving average technique. An illustrative example is also given for a practical data set.