Previous Time Series Research Articles

Online fuzzy time series analysis based on entropy discretization and a Fast Fourier Transform

Fuzzy time series analysis has been used successfully for forecasting in various domains including stock performance, academic enrollment, temperature, and traffic patterns. Research in this field has concentrated primarily on two issues: the reasonable partition of discourse, and defuzzification methods for discrete datasets. Both issues have a huge impact on the prediction performance of forecasting models. This paper integrates the entropy discretization technique with a Fast Fourier Transform (FFT) algorithm to develop a novel fuzzy time series forecasting model to resolve these issues. The Taiwan Stock Exchange Capitalization Weighted Stock Index (TAIEX) and Dow-Jones Industrial Average (DJIA) financial datasets were used to evaluate the model's performance. The results demonstrate that the presented model is a major improvement over previous fuzzy time series models produced by Chen (1996), Yu (2005), Chang et al. (2011), and Hsieh et al. (2011), and five other conventional time series models. The proposed model is implemented using the bootstrapping method, after which it incrementally updates its prediction capability. Results show that the proposed model's incremental learning mechanism allows it to effectively handle large online financial datasets.

INTEGRAL/SPI data segmentation to retrieve source intensity variations

The INTEGRAL/SPI, X-gamma-ray spectrometer (20 keV - 8 MeV) is an instrument for which recovering source intensity variations is not straightforward and can constitute a difficulty for data analysis. In most cases, determining the source intensity changes between exposures is largely based on a priori information. We propose techniques that help to overcome the difficulty related to source intensity variations, which make this step more rational. In addition, the constructed "synthetic" light curves should permit us to obtain a sky model that describes the data better and optimizes the source signal-to-noise ratios. For this purpose, the time intensity variation of each source was modeled as a combination of piecewise segments of time during which a given source exhibits a constant intensity. To optimize the signal-to-noise ratios, the number of segments was minimized. We present a first method that takes advantage of previous time series that can be obtained from another instrument on-board the INTEGRAL observatory. A data segmentation algorithm was then used to synthesize the time series into segments. The second method no longer needs external light curves, but solely SPI raw data. For this, we developed a specific algorithm that involves the SPI transfer function. The time segmentation algorithms that were developed solve a difficulty inherent to the SPI instrument, which is the intensity variations of sources between exposures, and it allows us to obtain more information about the sources' behavior.

An empirical test of exogenous versus endogenous growth models for the G-7 countries

One of the key differences between exogenous and endogenous growth models is that a transitory shock to investment share exhibits different long-run effects on per-capita output. Exploring this difference, the present paper evaluates the empirical relevance of the two growth models for the G-7 countries. The underlying shocks are identified by an application of a dynamic factor model. Results show that a transitory shock to investment share permanently increases per-capita output in four countries, offering support to the endogenous growth model. This shock also contributes considerably to accounting for the long-run variability of per-capita output. Overall, the endogenous model is found to be empirically more plausible than previous time series studies suggest.

A new improved forecasting method integrated fuzzy time series with the exponential smoothing method

This paper presents a new method of integrated fuzzy time series with the exponential smoothing method to forecast university enrolments. The data of historical enrolments of the University of Alabama shown in Liu et al. (2011) are adopted to illustrate the forecasting process of the proposed method. A comparison has been made with five previous fuzzy time series models. Meanwhile, the mean squared error has also been calculated as the evaluation criterion to illustrate the performance of the proposed method. The empirical analysis shows that the proposed model reflects the fluctuations in fuzzy time series better and provides better overall forecasting results than the five listed previous models.

Korean National Income Based on a Chain Index: 1953~2010

Korea’s national income statistics have been compiled by the Bank of Korea since 1953. However, there is a break in the time series. The current time series (1970 onward) is based on the ‘1993 SNA (System of National Accounts)’ suggested by the UN, and the previous time series (1953~1970) was based on the ‘1953 SNA’. The difference between the previous and current time series is 4.8% in 1970 when the two series overlap. The difference is even greater in terms of comparisons across industries. In addition, it has now become even more difficult to connect the current and the previous time series because, in 2009, the Bank of Korea introduced a chain weighted method for calculating the current time series (1970 onward). Under the chain weighted method, the time series underwent substantial modification; for instance, the economic growth rate during 1970~2005 is 0.9%p higher than the rate under the general method. This paper applies chain weighted values and the ‘1993 SNA’ to the previous time series (1953~1970) by utilizing various national account manuals published by the UN and previous Korean input-output tables in order to calculate a long term time series from 1953 to 2010 based on the same criteria as the current time series (1970 onward). In the revised time series, it appears that 1953 GDP at current basic prices is 3.5% higher and the growth rate for the period of 1953~1970 is 1.5%p higher each year than under the previous time series. Under the revised time series the size of the Korean economy as of 2010 is 50-fold bigger than that of 1953. In terms of industries, manufacturing and SOC show significant expansion whereas the extent of that of the service industry is relatively small.

Autoregressive integrated moving average model in food poisoning prediction in Hunan Province

To determine the application of autoregressive integrated moving average (ARIMA) model in food poisoning prediction in Hunan Province, and to provide scientific basis for the prevention and control of food poisoning. We collected the number of food poisoning from January 2003 to December 2009 in Hunan Province for ARIMA model fitting, and used food poisoning data of 2010 to verify the effect of model prediction. We predicted the number of food poisoning in 2011. ARIMA (0,1,1) (0,1,1)12 better fit the trends of the poisoning number in previous time periods and series, with prediction fitting error of 9.59%. The number of food poisoning in Hunan Province in 2011 was predicted to be 834. ARIMA model can better fit the number of food poisoning in the short term trends and series. If used for long-term forecasts.

Algorithms for computing spike time distance and point process prototypes with application to feline neuronal responses to acoustic stimuli

The spike-time distance metric and the prototype of a collection of point patterns can be used to provide a metric description of repeated observations of point processes. Due to algorithmic limitations in computing spike-time distance, previous work in determining approximations of the prototype has largely been limited to single dimensional or small data sets. We develop new algorithms for each of these methods that are suitable for larger data sets. The first, an incremental matching algorithm, is a method to compute spike-time distance. The second algorithm involves the use of kernel smoothing for prototype construction. These methods readily extend to multiple dimensions and to marked point processes. We use a combination of these approaches to analyze neuronal spike data of cats in different behavioral states and across animals, and we compare our results to previous time series analyses that used averaged frequency histograms. Evidence is found for short latency responses to the acoustic CS that were sensitive to behavioral state.

Fuzzy Multiple-Periods Time-Series Model Based on GSP and AR for Forecasting Gold Price

In the recent years, traditional time series model has been widely researched. The previous time series methods can predict future problems based on historical data, but have a problem that determines subjectively the length of intervals. Song and Chissom[6-7]proposed the fuzzy time series to solve the problem of traditional time series methods. So far, many researchers have proposed different fuzzy time series models to deal with uncertain and vague data. Besides, the consideration of a forecasting stage only discusses the relations for previous period and next period. In addition, a shortcoming of previous time series models didn’t consider appropriately the weights of fuzzy relations. This study builds fuzzy rule based on association rules and compute the cardinality of each fuzzy relation. Then, calculating the weights of fuzzy relations solve above problems. Moreover, the proposed method is able to build the multiple periods fuzzy rules based on concept of large itemsets of Apriori. To verify the proposed model, the gold price datasets is employed as experimental datasets. This study compares the forecasting accuracy of proposed model with other methods, and the comparison results show that the proposed method has better performance than other methods.

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

Abstract 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.

Including viral infection data supports an association between particulate pollution and respiratory admissions

To refine and revise previous air pollution, climate and health time series analysis in Christchurch, New Zealand, introducing viral identification data (positive identification count and outbreak, defined as two of more positive tests). The effects on daily respiratory admissions for five years (1998-2002) of air pollution (PM(10) ), climate and virology (incorporating actual counts and outbreaks of influenza A and B (INF), para influenza virus type 3 (PIV) and respiratory syncytial virus (RSV) were examined using generalised additive models (GAMs), which are one of semiparametric models. Results were also compared with a model that included climate and air pollution parameters but without the inclusion of virology data. The data were analysed aggregately and then stratified by age group and season. Different virology data detected various association levels. The highest estimates were a 3.93% (CI: 2.69-5.17) and a 3.88% (CI: 2.65-5.12) rise in respiratory admissions for a rise of 10 µg/m(3) annual PM(10) with outbreak and actual counts of PIV respectively for 0-19 years old with a three-day lag. Refining a statistical model with the addition of virology data gives a similar estimation of the association between PM(10) levels and respiratory admissions to previous research. Use of the indicator of an outbreak of viral infection appears to be similar to actual count of viruses detected.

Coning Algorithm Design by Explicit Frequency Shaping

A new optimum approach is presented for strapdown coning algorithm design based on Explicit matching of desired response to expected coning input magnitude as a function of coning frequency. Unlike previous time or frequency Taylor series expansion techniques, the new method achieves optimization through minimum least-squares estimation over a user selected design frequency range. The error being minimized is the square of the weighted algorithm error, the weighting factor being user specified coning amplitude over frequency. This methodology allows the coning algorithm coefficients to be independently designed for each coning axis to achieve optimal balanced performance over the frequency range of expected coning inputs. Performance of the new algorithm design technique is evaluated by comparison with previous time and frequency-series approaches, in discrete and stochastic coning environments, and over a time based extreme dynamic maneuvering profile. A generalized method is presented for translating system performance requirements into expected coning amplitude versus frequency profiles for calculating Explicit algorithm coefficients, and for evaluating coning algorithm accuracy. The paper includes derivations of generalized formulas for calculating algorithm coefficients for previous time and frequency-series design approaches.

Reply to Haddock, S.H.D. Reconsidering evidence for potential climate‐related increases in jellyfish

Following the publication of our paper (Attrill et al. 2007), we became quickly aware of a couple of errors. We have subsequently been collaborating with Dr. Chris Lynam (Lynam et al. 2004, 2005) to bring together our two datasets, explore the common patterns within our data, and attempt to provide a consensus on how climate is affecting gelatinous plankton in the North Sea. During this reanalysis, two errors within the data were discovered, one involving a transcription error of a column of residuals during de-trended analysis, the other a major data entry error deep in the Continuous Plankton Recorder (CPR) database for sector B2. Here we present a revised version of table 1 from Attrill et al. (2007) to incorporate corrections to these transcription and data entry errors. These corrections alter some of the results in our original data table, mainly to increase and strengthen the number of significant relations we found (e.g., for sector B2 and whole sea area); all previous main results remain robustly significant. Following discussions with Dr. Lynam, two clarifications of statements made in Attrill et al. (2007) are also required. Page 482, Results, last line of first column: ‘‘There were no...robust, consistent relations between jellyfish frequency and any environmental variables for B and D… contrary to the findings of previous shorter time series (Lynam et al. 2005).’’ The Lynam et al. (2004, 2005) papers presented no data for the D sector and found no link in the B sector, contrary to our revised results. Page 482, Discussion, paragraph 1, last sentence: ‘‘… positive association … North of Scotland (Lynam et al. 2005) … does not appear to be maintained.’’ Our paper did not report on any data that covered Lynam et al.’s (2005) North of Scotland area so the statement is not directly supported, although their positive relation North of Scotland, when considered in conjunction with inflow, may agree with the C2 and B2 results of Attrill et al. (2007).

A multifractal approach for stock market inefficiency

In this paper, the multifractality degree in a collection of developed and emerging stock market indices is evaluated. Empirical results suggest that the multifractality degree can be used as a quantifier to characterize the stage of market development of world stock indices. We develop a model to test the relationship between the stage of market development and the multifractality degree and find robust evidence that the relationship is negative, i.e., higher multifractality is associated with a less developed market. Thus, an inefficiency ranking can be derived from multifractal analysis. Finally, a link with previous volatility time series results is established.

Deciphering the pulsations of G 29-38 with optical time series spectroscopy

We present optical time series spectroscopy of the pulsating white dwarf star G 29-38 taken at the Very Large Telescope (VLT). By measuring the variations in brightness, Doppler shift and line shape of each spectrum, we explore the physics of pulsation and measure the spherical degree (l) of each stellar pulsation mode. We measure the physical motion of the g modes correlated with the brightness variations for three of the eight pulsation modes in this data set. The varying line shape reveals the spherical degree of the pulsations, an important quantity for properly modelling the interior of the star with asteroseismology. Performing fits to the Hβ, Hy and Hδ lines, we quantify the changing shape of the line and compare them to models and previous time series spectroscopy of G 29-38. These VLT data confirm several l identifications and add four new values, including an additional l = 2 and a possible l = 4. In total, from both sets of spectroscopy of G 29-38, eleven modes now have known spherical degrees.

A Parametric Study of Cognitive Defusion and the Believability and Discomfort of Negative Self-Relevant Thoughts

A previous time-series study showed that rapidly repeating a single-word version of a negative self-referential thought reduced the discomfort and the believability associated with that thought. The present parametric study examined whether durations of word repetition were differentially effective in altering the discomfort and believability of negative self-referential thought. In two studies, both discomfort and believability varied systematically with the duration of word repetition. The effects of rapid repetition on emotional discomfort bottomed out after 3 s to 10 s of rapid repetition, whereas the effects on believability did so after 20 s to 30 s of repetition. This study lends support to the cognitive defusion interpretation of the effect of word repetition, suggesting that emotional discomfort and believability may be distinctive functional aspects of cognitive events.

Farmland prices, structural breaks and panel data

Previous time series evidence has indicated that farmland prices and cash rents are not cointegrated, a finding at odds with the present value model of farmland prices. We argue that this failure to find cointegration may be due to low power of tests and to the presence of structural change representing a shifting risk premium on farmland investments. To accommodate this possibility, we use panel unit root and cointegration methods that are more powerful than conventional time series methods and allow for breaks in the cointegration relationship. Our results, based on a large panel covering 31 US states between 1960 and 2000, suggest that the present value model of farmland prices cannot be rejected.

AN ECONOMIC INTERPRETATION OF SUICIDE CYCLES IN JAPAN

Suicide rates in Japan have increased dramatically in recent years, making Japan’s male rate the highest among developed economies. This study revises the standard economic model of suicide to accommodate Japan’s experience, focusing on the change in human capital for the unemployed. We then use the new model and detrended data to empirically investigate the relationship between the suicide cycle and the unemployment cycle. Unlike previous aggregate time series studies, we find that the relationship between the suicide rate and the unemployment rate is significantly and robustly positive for both men and women even after controlling for several social variables. (JEL I12, J60, E30)

Modeling emotional content of music using system identification

Research was conducted to develop a methodology to model the emotional content of music as a function of time and musical features. Emotion is quantified using the dimensions valence and arousal, and system-identification techniques are used to create the models. Results demonstrate that system identification provides a means to generalize the emotional content for a genre of music. The average R2 statistic of a valid linear model structure is 21.9% for valence and 78.4% for arousal. The proposed method of constructing models of emotional content generalizes previous time-series models and removes ambiguity from classifiers of emotion.

An atmosphere–ocean time series model of global climate change

Time series models of global climate change tend to estimate a low climate-sensitivity (equilibrium effect on global temperature of doubling carbon dioxide concentrations) and a fast adjustment rate to equilibrium. These results may be biased by omission of a key variable—heat stored in the ocean. A time series model of the atmosphere–ocean climate system is developed, in which surface temperature (atmospheric temperature over land and sea surface temperature) moves towards a long-run equilibrium with both radiative forcing and ocean heat content, while ocean heat content accumulates the deviations from atmospheric equilibrium. This model is closely related to Granger and Lee's multicointegration model. As there are only 55 years of observations on ocean heat content, the Kalman filter is used to estimate heat content as a latent state variable, which is constrained by the available observations. This method could be applied to other climate change problems where there are only limited observations on key variables. The final model adopted relates surface temperature to the heat content of the upper 300 m of the ocean. The resulting parameter estimates are closer to theoretically expected values than those of previous time series models and the estimated climate sensitivity to a doubling of carbon dioxide is 4.4 K.

Meso-/Cenozoic basin and carbonate platform development in the SW-Dolomites unraveled by basin modelling and apatite FT analysis: Rosengarten and Latemar (Northern Italy)

The Rosengarten and Latemar Middle Triassic carbonate platforms indicate similar subsidence patterns during equivalent times of platform evolution as calculated by numerical basin reverse modelling. In the case of Rosengarten and Latemar, thermo-tectonic subsidence is the most significant among all three subsidence components (thermo-tectonic s., flexural s., compaction-induced s.). Hence, the influence of differential tectonics–as recorded in the formation of structural highs and lows in the Dolomites during the Lower to Middle Anisian–almost ceases in Late Anisian times. Basin modelling of Rosengarten and Latemar requires the correlation and integration of two sets of chronostratigraphic data from Buchenstein Fm (basin) and Schlern Dolomite Fm 1 (platform). Subsidence modelling argues for a microcycle (i.e. a single shallowing-upward cycle) duration of less than 1.97 ka in the cyclic succession at Latemar. However, previous time series analyses identify Milankovitch-cyclicities in larger-order cycle stacking patterns and thus indicate a microcycle duration of 4.2 ka. According to each model, total subsidence reaches values of 650–700 m/Ma (4.2 ka model) or 780–850 m/Ma (1.97 ka model) during the first stage of platform evolution. As both platforms successfully keep up with this subsidence peak, it is evident that the carbonate factory–despite its low diversity–must have recovered from the P/T faunal crisis. After this regionally observed peak, subsidence drops and triggers fast progradation at Rosengarten. Subsidence in the study area is controlled by active strike-slip tectonics and the subsequent formation of a magmatic chamber (Predazzo/Monzoni) above the deep-reaching Truden/Stava line. The area close to this tectonic line and to the volcanic centre of Predazzo/Monzoni undergoes a higher heat flow pulse during the Late Ladinian and is possibly buried at shallower levels than the Rosengarten area. As shown by fission tracks in apatites, the heat flow pulse almost disappears laterally underlining the localised thermal influence of the Late Ladinian volcano-thermal event of Predazzo/Monzoni.