Forecast combinations of computational intelligence and linear models for the NN5 time series forecasting competition

In this work we introduce the forecasting model with which we participated in the NN5 forecasting competition (the forecasting of 111 time series representing daily cash withdrawal amounts at ATM machines). The main idea of this model is to utilize the concept of forecast combination, which has proven to be an effective methodology in the forecasting literature. In the proposed system we attempted to follow a principled approach, and make use of some of the guidelines and concepts that are known in the forecasting literature to lead to superior performance. For example, we considered various previous comparison studies and time series competitions as guidance in determining which individual forecasting models to test (for possible inclusion in the forecast combination system). The final model ended up consisting of neural networks, Gaussian process regression, and linear models, combined by simple average. We also paid extra attention to the seasonality aspect, decomposing the seasonality into weekly (which is the strongest one), day of the month, and month of the year seasonality.     

Tourism,terrorism and political violence in Tunisia: Evidence from Markov-switching models

This study investigates the impact of terrorist attacks and political violence on the number of tourist arrivals and overnight stays in Tunisia. The dataset employed consists of monthly data that covers the period from January 2000 to September 2016, which includes several political and terrorist attacks in Tunisia and the region. Empirically, we investigate the true data generating process (DGP) of these two proxies of tourism activity by accounting for four statistical properties that characterize these series: (1) seasonality, (2) unit roots, (3) breaks, and (4) long memory behavior.Our empirical findings show strong evidence of stationarity, five breaks in the tourist arrival time series and spurious long memory behavior. By estimating a 3-state Markov switching model consisting of the mean, trend, and variance, we find that the Tunisian Jasmine revolution and two recent terrorist attacks, one at the Bardo National Museum on March 18, 2015 and the other at the tourist resort at Port El Kantaoui, Sousse on June 26, 2015, played an important role in influencing the tourism activity of the country. Our empirical findings show also that local shocks have a more important impact than international shocks in influencing tourism activity. Interestingly, we find that the effects of terrorist shocks have a long duration compared to political violence shocks. Several security, marketing, and economic policies have been proposed and discussed in the paper.     

Determinants of revenue per available room: Influential roles of average daily rate,demand, seasonality and yearly trend

Determining price per room to be charged to customers is an important decision to be taken by hotel management. Hotels frequently change their room rates based on the demand of room, occupancy rate, seasonal pattern, and strategies undertaken by other hotels on pricing. We formulated four models to analyse how various influencing variables, such as hotel price, demand, yearly trend and monthly seasonality influence hotel revenue per available room (RevPar). To analyse a case, we used monthly accommodation statistics for Sweden taken for Swedish Agency for Economic and Regional Growth and Statistics from January 2008 to July 2017. We carried out data analysis using both multiple regression and Multivariate Adaptive Regression Splines (MARS) model and found that application of MARS can help establishing a nonlinear relationship of RevPar with other determining variables in a superior way. We also proposed the possibility of developing a better forecasting model using MARS.     

US stock returns: are there seasons of excesses?

This article explores the existence of seasonality in the tails of stock returns. We use a parametric model to describe the returns, and obtain a proxy of the innovation distribution via a pre-processing model. Then, we develop a change-point algorithm capturing changes in the tails of the innovations. We confirm the good performance of the procedure through extensive Monte Carlo experiments. An empirical investigation using US stocks data shows that while the lower tail of the innovations is approximately constant over the year, the upper tail is larger in Winter than in Summer, in 9 out of 12 industries.     

The treatment of seasonality in error correction models as unobserved component: a case study for an Austrian consumption function

This paper deals with an alternative approach to treating seasonality in error correction models for consumption with a parsimonious parameterization as proposed by Harvey and Scott. We introduce an unobserved seasonal component into an error correction model for Austrian consumer expenditures on nondurables and services and compare the results with different approaches. The use of stochastic seasonal results in a definite improvement of the estimated model. First version received: October 1997/Final version received: May 2000     

An unequal burden: Intra-household dimensions of seasonal health in Tanzania

Seasonal variation in nutritional status is a concern in sub-Saharan Africa. Seasonality in the food consumption of young Tanzanian children has a substantial and negative impact on later life outcomes. Seasonal variation in adult nutritional status has implications for labor productivity and prenatal health. Just as mean poverty status or mean health status varies within households, seasonal patterns in health status are likely to vary within households, as well as the mechanisms underlying seasonal variation. We parameterize and compare seasonality in nutritional status across multiple types of household members in rural, farming households in Tanzania, using a novel anthropometric measure of body mass index z-score that is comparable across adults and children. Young children not yet in school and working adults are most vulnerable to seasonal fluctuations in nutritional status. Children in school and older adults are relatively shielded. Seasonal variation in the nutritional status of working adults can be partly explained by variation in dietary quality and agricultural labor hours. Seasonal variation in the nutritional status of young children is not explained by either factor, nor is it mitigated by market access. Our results suggest we do not understand the data generating process behind seasonality in the nutritional status of young children, despite the critical implications of this seasonality for development and later life productivity.     

Considerations of a retail forecasting practitioner

Forecasts can be used in an extraordinarily diverse range of ways across many domains in which forecasting practitioners work continuously towards improving their forecasts. Each of these domains may require the analysis of different kinds of inputs and special considerations. Even within a given domain, such as retail, there may be many similar use cases of the same kind of forecast, which can lead to practitioners making different decisions. This paper discusses several of the important decision points that practitioners must work through and uses item-level sales forecasting in the retail domain as leveraged by pricing and inventory management as examples of the different paths that may be taken. It considers how each use can lead to a different forecasting objective, and a corresponding focus on different error metrics. In addition, there are several tradeoffs in the forecasting methods that are used to meet each of the objectives best, including the kinds of models used, the running time speed, and forecast accuracy requirements.     

An examination of the seasonality in industrial real estate construction

Summary

Though some work has been done to identify variables affecting seasonal variation in detached housing, almost nothing has been done to identify the factors affecting seasonal variations in industrial real estate construction. This paper attempts to identify some of these seasonal factors and their influence on industrial real estate construction. A Chow Test shows that there is a significant difference in the relationship between the amount of industrial real estate put in place and specific economic variables.     

Time zone normalization of FX seasonality

Abstract

This paper provides empirical evidence that the particular intra-day seasonality observed in the Foreign Exchange market is indeed due to the different geographical locations of its traders. Analysing more than 2 years of real transactions from a microscopic perspective, we design a procedure that accounts for the time zones from which traders operate. The resulting normalized intra-day seasonality shows a pattern akin to those observed in regulated exchanges where traders are more active at the beginning and at the end of their session.     

An application of rule-based forecasting to a situation lacking domain knowledge

Rule-based forecasting (RBF) uses rules to combine forecasts from simple extrapolation methods. Weights for combining the rules use statistical and domain-based features of time series. RBF was originally developed, tested, and validated only on annual data. For the M3-Competition, three major modifications were made to RBF. First, due to the absence of much in the way of domain knowledge, we prepared the forecasts under the assumption that no domain knowledge was available. This removes what we believe is one of RBF’s primary advantages. We had to re-calibrate some of the rules relating to causal forces to allow for this lack of domain knowledge. Second, automatic identification procedures were used for six time-series features that had previously been identified using judgment. This was done to reduce cost and improve reliability. Third, we simplified the rule-base by removing one method from the four that were used in the original implementation. Although this resulted in some loss in accuracy, it reduced the number of rules in the rule-base from 99 to 64. This version of RBF still benefits from the use of prior findings on extrapolation, so we expected that it would be substantially more accurate than the random walk and somewhat more accurate than equal weights combining. Because most of the previous work on RBF was done using annual data, we especially expected it to perform well with annual data.