Previous researches show that buy (growth) companies conduct income increasing earnings management in order to meet forecasts and generate positive forecast Errors (FEs). This behavior however, is not inherent in sell (non-growth) companies. Using the aforementioned background, this research hypothesizes that since sell companies are pressured to avoid income increasing earnings management, they are capable, and in fact more inclined, to pursue income decreasing Forecast Management (FM) with the purpose of generating positive FEs. Using a sample of 6553 firm-years of companies that are listed in the NYSE between the years 2005-2010, the study determines that sell companies conduct income decreasing FM to generate positive FEs. However, the frequency of positive FEs of sell companies does not exceed that of buy companies. Using the efficiency perspective, the study suggests that even though buy and sell companies have immense motivation in avoiding negative FEs, they exploit different but efficient strategies, respectively, in order to meet forecasts. Furthermore, the findings illuminated the complexities behind informative and opportunistic forecasts that falls under the efficiency versus opportunistic theories in literature.
Vegetation fires have become an increasing problem in tropical environments as a consequence of socioeconomic pressures and subsequent land-use change. In response, fire management systems are being developed. This study set out to determine the relationships between two aspects of the fire problems in western Indonesia and Malaysia, and two components of the Canadian Forest Fire Weather Index System. The study resulted in a new method for calibrating components of fire danger rating systems based on satellite fire detection (hotspot) data. Once the climate was accounted for, a problematic number of fires were related to high levels of the Fine Fuel Moisture Code. The relationship between climate, Fine Fuel Moisture Code, and hotspot occurrence was used to calibrate Fire Occurrence Potential classes where low accounted for 3% of the fires from 1994 to 2000, moderate accounted for 25%, high 26%, and extreme 38%. Further problems arise when there are large clusters of fires burning that may consume valuable land or produce local smoke pollution. Once the climate was taken into account, the hotspot load (number and size of clusters of hotspots) was related to the Fire Weather Index. The relationship between climate, Fire Weather Index, and hotspot load was used to calibrate Fire Load Potential classes. Low Fire Load Potential conditions (75% of an average year) corresponded with 24% of the hotspot clusters, which had an average size of 30% of the largest cluster. In contrast, extreme Fire Load Potential conditions (1% of an average year) corresponded with 30% of the hotspot clusters, which had an average size of 58% of the maximum. Both Fire Occurrence Potential and Fire Load Potential calibrations were successfully validated with data from 2001. This study showed that when ground measurements are not available, fire statistics derived from satellite fire detection archives can be reliably used for calibration. More importantly, as a result of this work, Malaysia and Indonesia have two new sources of information to initiate fire prevention and suppression activities.
Time series forecasting has gained much attention due to its many practical applications. Higher-order neural network with recurrent feedback is a powerful technique that has been used successfully for time series forecasting. It maintains fast learning and the ability to learn the dynamics of the time series over time. Network output feedback is the most common recurrent feedback for many recurrent neural network models. However, not much attention has been paid to the use of network error feedback instead of network output feedback. In this study, we propose a novel model, called Ridge Polynomial Neural Network with Error Feedback (RPNN-EF) that incorporates higher order terms, recurrence and error feedback. To evaluate the performance of RPNN-EF, we used four univariate time series with different forecasting horizons, namely star brightness, monthly smoothed sunspot numbers, daily Euro/Dollar exchange rate, and Mackey-Glass time-delay differential equation. We compared the forecasting performance of RPNN-EF with the ordinary Ridge Polynomial Neural Network (RPNN) and the Dynamic Ridge Polynomial Neural Network (DRPNN). Simulation results showed an average 23.34% improvement in Root Mean Square Error (RMSE) with respect to RPNN and an average 10.74% improvement with respect to DRPNN. That means that using network errors during training helps enhance the overall forecasting performance for the network.
This study applies the clonal selection algorithm (CSA) in an artificial immune system (AIS) as an alternative method to predicting future rainfall data. The stochastic and the artificial neural network techniques are commonly used in hydrology. However, in this study a novel technique for forecasting rainfall was established. Results from this study have proven that the theory of biological immune systems could be technically applied to time series data. Biological immune systems are nonlinear and chaotic in nature similar to the daily rainfall data. This study discovered that the proposed CSA was able to predict the daily rainfall data with an accuracy of 90% during the model training stage. In the testing stage, the results showed that an accuracy between the actual and the generated data was within the range of 75 to 92%. Thus, the CSA approach shows a new method in rainfall data prediction.
This study presents the use of a wavelet-based time series model to forecast the daily average particulate matter with an aerodynamic diameter of less than 10 μm (PM10) in Peninsular Malaysia. The highlight of this study is the use of a discrete wavelet transform (DWT) in order to improve the forecast accuracy. The DWT was applied to convert the highly variable PM10 series into more stable approximations and details sub-series, and the ARIMA-GARCH time series models were developed for each sub-series. Two different forecast periods, one was during normal days, while the other was during haze episodes, were designed to justify the usefulness of DWT. The models' performance was evaluated by four indices, namely root mean square error, mean absolute percentage error, probability of detection and false alarm rate. The results showed that the model incorporated with DWT yielded more accurate forecasts than the conventional method without DWT for both the forecast periods, and the improvement was more prominent for the period during the haze episodes.
Breast cancer is the most common cancer among women and it is one of the main causes of death for women worldwide. To attain an optimum medical treatment for breast cancer, an early breast cancer detection is crucial. This paper proposes a multi- stage feature selection method that extracts statistically significant features for breast cancer size detection using proposed data normalization techniques. Ultra-wideband (UWB) signals, controlled using microcontroller are transmitted via an antenna from one end of the breast phantom and are received on the other end. These ultra-wideband analogue signals are represented in both time and frequency domain. The preprocessed digital data is passed to the proposed multi- stage feature selection algorithm. This algorithm has four selection stages. It comprises of data normalization methods, feature extraction, data dimensional reduction and feature fusion. The output data is fused together to form the proposed datasets, namely, 8-HybridFeature, 9-HybridFeature and 10-HybridFeature datasets. The classification performance of these datasets is tested using the Support Vector Machine, Probabilistic Neural Network and Naïve Bayes classifiers for breast cancer size classification. The research findings indicate that the 8-HybridFeature dataset performs better in comparison to the other two datasets. For the 8-HybridFeature dataset, the Naïve Bayes classifier (91.98%) outperformed the Support Vector Machine (90.44%) and Probabilistic Neural Network (80.05%) classifiers in terms of classification accuracy. The finalized method is tested and visualized in the MATLAB based 2D and 3D environment.
The increasing trend in the number new cases of influenza every year as reported by WHO is concerning, especially in Malaysia. To date, there is no local research under healthcare sector that implements the time series forecasting methods to predict future disease outbreak in Malaysia, specifically influenza. Addressing the problem could increase awareness of the disease and could help healthcare workers to be more prepared in preventing the widespread of the disease. This paper intends to perform a hybrid ARIMA-SVR approach in forecasting monthly influenza cases in Malaysia. Autoregressive Integrated Moving Average (ARIMA) model (using Box-Jenkins method) and Support Vector Regression (SVR) model were used to capture the linear and nonlinear components in the monthly influenza cases, respectively. It was forecasted that the performance of the hybrid model would improve. The data from World Health Organization (WHO) websites consisting of weekly Influenza Serology A cases in Malaysia from the year 2006 until 2019 have been used for this study. The data were recategorized into monthly data. The findings of the study showed that the monthly influenza cases could be efficiently forecasted using three comparator models as all models outperformed the benchmark model (Naïve model). However, SVR with linear kernel produced the lowest values of RMSE and MAE for the test dataset suggesting the best performance out of the other comparators. This suggested that SVR has the potential to produce more consistent results in forecasting future values when compared with ARIMA and the ARIMA-SVR hybrid model.
The annual disease incident worldwide is desirable to be predicted for taking appropriate policy to prevent disease outbreak. This chapter considers the performance of different forecasting method to predict the future number of disease incidence, especially for seasonal disease. Six forecasting methods, namely linear regression, moving average, decomposition, Holt-Winter's, ARIMA, and artificial neural network (ANN), were used for disease forecasting on tuberculosis monthly data. The model derived met the requirement of time series with seasonality pattern and downward trend. The forecasting performance was compared using similar error measure in the base of the last 5 years forecast result. The findings indicate that ARIMA model was the most appropriate model since it obtained the less relatively error than the other model.
We used the global fire detection record provided by the satellite-based Moderate Resolution Imaging Spectroradiometer (MODIS) to determine the number of fires detected inside 823 tropical and subtropical moist forest reserves and for contiguous buffer areas 5, 10, and 15 km wide. The ratio of fire detection densities (detections per square kilometer) inside reserves to their contiguous buffer areas provided an index of reserve effectiveness. Fire detection density was significantly lower inside reserves than in paired, contiguous buffer areas but varied by five orders of magnitude among reserves. The buffer: reserve detection ratio varied by up to four orders of magnitude among reserves within a single country, and median values varied by three orders of magnitude among countries. Reserves tended to be least effective at reducing fire frequency in many poorer countries and in countries beset by corruption. Countries with the most successful reserves include Costa Rica, Jamaica, Malaysia, and Taiwan and the Indonesian island of Java. Countries with the most problematic reserves include Cambodia, Guatemala, Paraguay, and Sierra Leone and the Indonesian portion of Borneo. We provide fire detection density for 3964 tropical and subtropical reserves and their buffer areas in the hope that these data will expedite further analyses that might lead to improved management of tropical reserves.
Dengue fever is a mosquito-borne disease that affects nearly 3.9 billion people globally. Dengue remains endemic in Malaysia since its outbreak in the 1980's, with its highest concentration of cases in the state of Selangor. Predictors of dengue fever outbreaks could provide timely information for health officials to implement preventative actions. In this study, five districts in Selangor, Malaysia, that demonstrated the highest incidence of dengue fever from 2013 to 2017 were evaluated for the best machine learning model to predict Dengue outbreaks. Climate variables such as temperature, wind speed, humidity and rainfall were used in each model. Based on results, the SVM (linear kernel) exhibited the best prediction performance (Accuracy = 70%, Sensitivity = 14%, Specificity = 95%, Precision = 56%). However, the sensitivity for SVM (linear) for the testing sample increased up to 63.54% compared to 14.4% for imbalanced data (original data). The week-of-the-year was the most important predictor in the SVM model. This study exemplifies that machine learning has respectable potential for the prediction of dengue outbreaks. Future research should consider boosting, or using, nature inspired algorithms to develop a dengue prediction model.
Malaria remains a major health problem in Sudan. With a population exceeding 39 million, there are around 7.5 million cases and 35,000 deaths every year. The predicted distribution of malaria derived from climate factors such as maximum and minimum temperatures, rainfall and relative humidity was compared with the actual number of malaria cases in Sudan for the period 2004 to 2010. The predictive calculations were done by fuzzy logic suitability (FLS) applied to the numerical distribution of malaria transmission based on the life cycle characteristics of the Anopheles mosquito accounting for the impact of climate factors on malaria transmission. This information is visualized as a series of maps (presented in video format) using a geographical information systems (GIS) approach. The climate factors were found to be suitable for malaria transmission in the period of May to October, whereas the actual case rates of malaria were high from June to November indicating a positive correlation. While comparisons between the prediction model for June and the case rate model for July did not show a high degree of association (18%), the results later in the year were better, reaching the highest level (55%) for October prediction and November case rate.
k-nearest neighbour (k-NN) has been shown to be an effective learning algorithm for classification and prediction. However, the application of k-NN for prediction and classification in specific sport is still in its infancy. The present study classified and predicted high and low potential archers from a set of physical fitness variables trained on a variation of k-NN algorithms and logistic regression. 50 youth archers with the mean age and standard deviation of (17.0 ± 0.56) years drawn from various archery programmes completed a one end archery shooting score test. Standard fitness measurements of the handgrip, vertical jump, standing broad jump, static balance, upper muscle strength and the core muscle strength were conducted. Multiple linear regression was utilised to ascertain the significant variables that affect the shooting score. It was demonstrated from the analysis that core muscle strength and vertical jump were statistically significant. Hierarchical agglomerative cluster analysis (HACA) was used to cluster the archers based on the significant variables identified. k-NN model variations, i.e., fine, medium, coarse, cosine, cubic and weighted functions as well as logistic regression, were trained based on the significant performance variables. The HACA clustered the archers into high potential archers (HPA) and low potential archers (LPA). The weighted k-NN outperformed all the tested models at itdemonstrated reasonably good classification on the evaluated indicators with an accuracy of 82.5 ± 4.75% for the prediction of the HPA and the LPA. Moreover, the performance of the classifiers was further investigated against fresh data, which also indicates the efficacy of the weighted k-NN model. These findings could be valuable to coaches and sports managers to recognise high potential archers from a combination of the selected few physical fitness performance indicators identified which would subsequently save cost, time and energy for a talent identification programme.