Forecasting of groundwater level variations is a significantly needed in groundwater resource management. Precise water level prediction assists in practical and optimal usage of water resources. The main objective of using an artificial neural network (ANN) was to investigate the feasibility of feed-forward, Elman and Cascade forward neural networks with different algorithms to estimate groundwater levels in the Langat Basin from 2007 to 2013. In order to examine the accuracy of monthly water level forecasts, effectiveness of the steepness coefficient in the sigmoid function of a developed ANN model was evaluated in this research. The performance of the models was evaluated using the mean squared error (MSE) and the correlation coefficient (R). The results indicated that the ANN technique was well suited for forecasting groundwater levels. All models developed had shown acceptable results. Based on the observation, the feed-forward neural network model optimized with the Levenberg-Marquardt algorithms showed the most beneficial results with the minimum MSE value of (0.048) and maximum R value of (0.839), obtained for simulation of groundwater levels. The present research conclusively showed the capability of ANNs to provide excellent estimation accuracy and valuable sensitivity analyses.
A noticeable increase in drought frequency and severity has been observed across the globe due to climate change, which attracted scientists in development of drought prediction models for mitigation of impacts. Droughts are usually monitored using drought indices (DIs), most of which are probabilistic and therefore, highly stochastic and non-linear. The current research investigated the capability of different versions of relatively well-explored machine learning (ML) models including random forest (RF), minimum probability machine regression (MPMR), M5 Tree (M5tree), extreme learning machine (ELM) and online sequential-ELM (OSELM) in predicting the most widely used DI known as standardized precipitation index (SPI) at multiple month horizons (i.e., 1, 3, 6 and 12). Models were developed using monthly rainfall data for the period of 1949-2013 at four meteorological stations namely, Barisal, Bogra, Faridpur and Mymensingh, each representing a geographical region of Bangladesh which frequently experiences droughts. The model inputs were decided based on correlation statistics and the prediction capability was evaluated using several statistical metrics including mean square error (MSE), root mean square error (RMSE), mean absolute error (MAE), correlation coefficient (R), Willmott's Index of agreement (WI), Nash Sutcliffe efficiency (NSE), and Legates and McCabe Index (LM). The results revealed that the proposed models are reliable and robust in predicting droughts in the region. Comparison of the models revealed ELM as the best model in forecasting droughts with minimal RMSE in the range of 0.07-0.85, 0.08-0.76, 0.062-0.80 and 0.042-0.605 for Barisal, Bogra, Faridpur and Mymensingh, respectively for all the SPI scales except one-month SPI for which the RF showed the best performance with minimal RMSE of 0.57, 0.45, 0.59 and 0.42, respectively.
Dengue is hyper-endemic in Singapore and Malaysia, and daily movement rates between the two countries are consistently high, allowing inference on the role of local transmission and imported dengue cases. This paper describes a custom built sparse space-time autoregressive (SSTAR) model to infer and forecast contemporaneous and future dengue transmission patterns in Singapore and 16 administrative regions within Malaysia, taking into account connectivity and geographical adjacency between regions as well as climatic factors. A modification to forecast impulse responses is developed for the case of the SSTAR and is used to simulate changes in dengue transmission in neighbouring regions following a disturbance. The results indicate that there are long-term responses of the neighbouring regions to shocks in a region. By computation of variable inclusion probabilities, we found that each region's own past counts were important to describe contemporaneous case counts. In 15 out of 16 regions, other regions case counts were important to describe contemporaneous case counts even after controlling for past local dengue transmissions and exogenous factors. Leave-one-region-out analysis using SSTAR showed that dengue transmission counts could be reconstructed for 13 of 16 regions' counts using external dengue transmissions compared to a climate only approach. Lastly, one to four week ahead forecasts from the SSTAR were more accurate than baseline univariate autoregressions.
As we move towards the knowledge (K) era, the challenge in R&D is to focus on the development of original K-based products that can compete in the global market. The development of commercially viable, patented K-based products within a university environment require an innovation system and innovation policies in place and a change in the paradigm towards the approach to research. A crucial agent towards the success of the innovation system is development and training of the human capital that would be the future drivers of the K-industry. Awareness of intellectual property rights, the need for original research, entrepreneurship as well as the development and strengthening of self-confidence and leadership are among the factors needed towards the training of K-workers facing the new economy.
Iran is situated in a very diverse environmental area. The climate of the region is varied and influencedby different patterns. In order to best describe the expected climate change impacts for the region,climate change scenarios and climate variables must be developed on a regional, or even site-specific,scale. The weather generator is one of the valid downscaling methods. In the current study, LARSWG(a weather generator) and the outputs from ECHO-G for present climate, as well as future timeslice of 2010-2039 based on A1 scenario, were used to evaluate LARS-WG as a tool at 13 synopticstations located in the north and northeast parts of Iran. The results obtained in this study illustratethat LARS-WG has a reasonable capability of simulating the minimum and maximum temperaturesand precipitation. In addition, the results showed that the mean precipitation decreased in Semnan, thesouth of Khorasan and Golestan. Meanwhile, the mean temperature during 2010-2039 would increaseby 0.5°C, especially in the cold season.
An accurate forecasting of tropospheric ozone (O3) concentration is benefi-
cial for strategic planning of air quality. In this study, various forecasting techniques are
used to forecast the daily maximum O3 concentration levels at a monitoring station
in the Klang Valley, Malaysia. The Box-Jenkins autoregressive integrated movingaverage
(ARIMA) approach and three types of neural network models, namely, backpropagation
neural network, Elman recurrent neural network and radial basis function
neural network are considered. The daily maximum data, spanning from 1 January
2011 to 7 August 2011, was obtained from the Department of Environment, Malaysia.
The performance of the four methods in forecasting future values of ozone concentrations
is evaluated based on three criteria, which are root mean square error (RMSE),
mean absolute error (MAE) and mean absolute percentage error (MAPE). The findings
show that the Box-Jenkins approach outperformed the artificial neural network
methods.
Space weather forecasting and its importance for the power and communication industry have inspired research related to TEC forecasting lately. Research has attempted to establish an empirical model approach for TEC prediction. In this paper, artificial neural networks (ANNs) have been applied in total electron content using GPS Ionospheric Scintillation and TEC Monitor (GISTM) data from UKM Station. The TEC prediction will be useful in improving the quality of current GNSS applications, such as in automobiles, road mapping, location-based advertising, personal navigation or logistics. Hence, a neural network model was designed with relevant features and customised parameters. Various types of input data and data representations from the ionospheric activity were used for the chosen network structure, which was a three-layer perceptron trained by feed forward back propagation method and tested on the chosen test data. We found that the optimum RMSE occurred with 10 nodes as the best NN for GISTM UKM station for the studied period with RMSE 1.3457 TECU. An analysis was made to compare the TEC from the measured TEC with neural network prediction and from IRI-corr model. The results showed that the NN model forecast the TEC values close to the measured TEC values with 9.96% of relative error. Thus, the forecasting of total electron content has the potential to be implemented successfully with larger data set from multi-centred environment.
Impacts of climate change on the hydrologic processes under future climate change conditions were assessed over Muda and Dungun watersheds of Peninsular Malaysia by means of a coupled regional climate and physically-based hydrology model utilizing an ensemble of future climate change projections. An ensemble of 15 different future climate realizations from coarse resolution global climate models' (GCMs) projections for the 21st century was dynamically downscaled to 6km resolution over Peninsular Malaysia by a regional climate model, which was then coupled with the watershed hydrology model WEHY through the atmospheric boundary layer over Muda and Dungun watersheds. Hydrologic simulations were carried out at hourly increments and at hillslope-scale in order to assess the impacts of climate change on the water balances and flooding conditions in the 21st century. The coupled regional climate and hydrology model was simulated for a duration of 90years for each of the 15 realizations. It is demonstrated that the increase in mean monthly flows due to the impact of expected climate change during 2040-2100 is statistically significant from April to May and from July to October at Muda watershed. Also, the increase in mean monthly flows is shown to be significant in November during 2030-2070 and from November to December during 2070-2100 at Dungun watershed. In other words, the impact of the expected climate change will be significant during the northeast and southwest monsoon seasons at Muda watershed and during the northeast monsoon season at Dungun watershed. Furthermore, the flood frequency analyses for both watersheds indicated an overall increasing trend in the second half of the 21st century.
Interest in Indigenous Knowledge (IK) system has been particularly highlighted in
flood disasters, due to the likely increase of flood events resulting from
anthropogenic climate change through heavy precipitation, increased catchment
wetness, and sea level rise. Therefore, bringing IK of flood risk reduction into focus
and context to deepen the understanding of how people manage their own changing
circumstances can bring more pertinent information about flood risk reduction. This
paper reviews the significance of IK in flood risk reduction. Specifically, the paper
discusses IK flood forecasting, early warning signs, adaptation and coping strategies
in flood risk reduction around the world. The Methodological approach employed for
this paper is the review of existing literature on IK in flood Disaster Risk Reduction
(DRR), and then a summary of the outcomes of the studies reviewed was discussed.
However, it was deduced from the review undertaken, the need for an intensive
empirical study to be conducted to explore how efficient these strategies or
techniques are, in relation to flood risk reduction, which this paper strongly
recommends for further investigation. Additionally, the paper concludes by
emphasizing that although the IK of flood risk reduction is embedded in varied
regions around the globe, still there is a need for further study to be carried out in
order to unveil why the similarities and variations in flood risk reduction
practices/strategies between regions.
In this paper, an interactively recurrent functional neural fuzzy network (IRFNFN) with fuzzy differential evolution (FDE)
learning method was proposed for solving the control and the prediction problems. The traditional differential evolution
(DE) method easily gets trapped in a local optimum during the learning process, but the proposed fuzzy differential
evolution algorithm can overcome this shortcoming. Through the information sharing of nodes in the interactive layer,
the proposed IRFNFN can effectively reduce the number of required rule nodes and improve the overall performance of
the network. Finally, the IRFNFN model and associated FDE learning algorithm were applied to the control system of the
water bath temperature and the forecast of the sunspot number. The experimental results demonstrate the effectiveness
of the proposed method.
Investment funds are growing in Malaysia since people are more knowledgeable about
investments and aware of investment opportunities in order to secure good savings for the
future. These investments include unit trusts, gold, fixed deposits, stock prices and property
investments. It is essential for individuals or organisations to know the value of future share
prices of their investment portfolio in order to predict the profit or loss in the future. The
purpose of study is to identify the best duration of historical data and forecast days in order
to accurately forecast share prices. The study uses Geometric Brownian Motion model in
forecasting share prices of companies in Bursa Malaysia. This study focused on 40 listed
companies in Bursa Malaysia from the top gainers list. It was found that 65 historical days
could forecast the share prices for 21 days accurately.
Predicting the number of defects in software at the method level is important. However, little or no research has focused on method-level defect prediction. Therefore, considerable efforts are still required to demonstrate how method-level defect prediction can be achieved for a new software version. In the current study, we present an analysis of the relevant information obtained from the current version of a software product to construct regression models to predict the estimated number of defects in a new version using the variables of defect density, defect velocity and defect introduction time, which show considerable correlation with the number of method-level defects. These variables also show a mathematical relationship between defect density and defect acceleration at the method level, further indicating that the increase in the number of defects and the defect density are functions of the defect acceleration. We report an experiment conducted on the Finding Faults Using Ensemble Learners (ELFF) open-source Java projects, which contain 289,132 methods. The results show correlation coefficients of 60% for the defect density, -4% for the defect introduction time, and 93% for the defect velocity. These findings indicate that the average defect velocity shows a firm and considerable correlation with the number of defects at the method level. The proposed approach also motivates an investigation and comparison of the average performances of classifiers before and after method-level data preprocessing and of the level of entropy in the datasets.
The growing number of multi-population mortality models in the recent years signifies the mortality improvement in
developed countries. In this case, there exists a narrowing gap of sex-differential in life expectancy between populations;
hence multi-population mortality models are designed to assimilate the correlation between populations. The present
study considers two extensions of the single-population Lee-Carter model, namely the independent model and augmented
common factor model. The independent model incorporates the information between male and female separately
whereas the augmented common factor model incorporates the information between male and female simultaneously.
The methods are demonstrated in two perspectives: First is by applying them to Malaysian mortality data and second
is by comparing the significance of the methods to the annuity pricing. The performances of the two methods are then
compared in which has been found that the augmented common factor model is more superior in terms of historical fit,
forecast performance, and annuity pricing.
Road traffic accidents have been listed in the top 10 global causes of death for many decades. Traditional measures such as education and legislation have contributed to limited improvements in terms of reducing accidents due to people driving in undesirable statuses, such as when suffering from stress or drowsiness. Attention is drawn to predicting drivers' future status so that precautions can be taken in advance as effective preventative measures. Common prediction algorithms include recurrent neural networks (RNNs), gated recurrent units (GRUs), and long short-term memory (LSTM) networks. To benefit from the advantages of each algorithm, nondominated sorting genetic algorithm-III (NSGA-III) can be applied to merge the three algorithms. This is named NSGA-III-optimized RNN-GRU-LSTM. An analysis can be made to compare the proposed prediction algorithm with the individual RNN, GRU, and LSTM algorithms. Our proposed model improves the overall accuracy by 11.2-13.6% and 10.2-12.2% in driver stress prediction and driver drowsiness prediction, respectively. Likewise, it improves the overall accuracy by 6.9-12.7% and 6.9-8.9%, respectively, compared with boosting learning with multiple RNNs, multiple GRUs, and multiple LSTMs algorithms. Compared with existing works, this proposal offers to enhance performance by taking some key factors into account-namely, using a real-world driving dataset, a greater sample size, hybrid algorithms, and cross-validation. Future research directions have been suggested for further exploration and performance enhancement.
Fuzzy time series (FTS) forecasting models show a great performance in predicting time series, such as air pollution time series. However, they have caused major issues by utilizing random partitioning of the universe of discourse and ignoring repeated fuzzy sets. In this study, a novel hybrid forecasting model by integrating fuzzy time series to Markov chain and C-Means clustering techniques with an optimal number of clusters is presented. This hybridization contributes to generating effective lengths of intervals and thus, improving the model accuracy. The proposed model was verified and validated with real time series data sets, which are the benchmark data of actual trading of Taiwan Stock Exchange Capitalization Weighted Stock Index (TAIEX) and PM10 concentration data from Melaka, Malaysia. In addition, a comparison was made with some existing fuzzy time series models. Furthermore, the mean absolute percentage error, mean squared error and Theil's U statistic were calculated as evaluation criteria to illustrate the performance of the proposed model. The empirical analysis shows that the proposed model handles the time series data sets more efficiently and provides better overall forecasting results than existing FTS models. The results prove that the proposed model has greatly improved the prediction accuracy, for which it outperforms several fuzzy time series models. Therefore, it can be concluded that the proposed model is a better option for forecasting air pollution parameters and any kind of random parameters.
With many countries experiencing a resurgence in COVID-19 cases, it is important to forecast disease trends to enable effective planning and implementation of control measures. This study aims to develop Seasonal Autoregressive Integrated Moving Average (SARIMA) models using 593 data points and smoothened case and covariate time-series data to generate a 28-day forecast of COVID-19 case trends during the third wave in Malaysia. SARIMA models were developed using COVID-19 case data sourced from the Ministry of Health Malaysia's official website. Model training and validation was conducted from 22 January 2020 to 5 September 2021 using daily COVID-19 case data. The SARIMA model with the lowest root mean square error (RMSE), mean absolute percentage error (MAE) and Bayesian information criterion (BIC) was selected to generate forecasts from 6 September to 3 October 2021. The best SARIMA model with a RMSE = 73.374, MAE = 39.716 and BIC = 8.656 showed a downward trend of COVID-19 cases during the forecast period, wherein the observed daily cases were within the forecast range. The majority (89%) of the difference between the forecasted and observed values was well within a deviation range of 25%. Based on this work, we conclude that SARIMA models developed in this paper using 593 data points and smoothened data and sensitive covariates can generate accurate forecast of COVID-19 case trends.
Gross domestic product (GDP) is an important indicator for determining a country's or region's economic status and development level, and it is closely linked to inflation, unemployment, and economic growth rates. These basic indicators can comprehensively and effectively reflect a country's or region's future economic development. The center of radial basis function neural network and smoothing factor to take a uniform distribution of the random radial basis function artificial neural network will be the focus of this study. This stochastic learning method is a useful addition to the existing methods for determining the center and smoothing factors of radial basis function neural networks, and it can also help the network more efficiently train. GDP forecasting is aided by the genetic algorithm radial basis neural network, which allows the government to make timely and effective macrocontrol plans based on the forecast trend of GDP in the region. This study uses the genetic algorithm radial basis, neural network model, to make judgments on the relationships contained in this sequence and compare and analyze the prediction effect and generalization ability of the model to verify the applicability of the genetic algorithm radial basis, neural network model, based on the modeling of historical data, which may contain linear and nonlinear relationships by itself, so this study uses the genetic algorithm radial basis, neural network model, to make, compare, and analyze judgments on the relationships contained in this sequence.
Combining forecast values based on simple univariate models may produce more favourable results than complex models. In this study, the results of combining the forecast values of Naïve model, Single Exponential Smoothing Model, The Autoregressive Moving Average (ARIMA) model, and Holt Method are shown to be superior to that of the Error Correction Model (ECM).Malaysia’s unemployment rates data are used in this study. The independent variable used in the ECM formulation is the industrial production index. Both data sets were collected for the months of January 2004 to December 2010. The selection criteria used to determine the best model, is the Mean Square Error (MSE), Root Mean Squared Error (RMSE) and Mean Absolute Percentage Error (MAPE). Initial findings showed that both time series data sets were not influenced by the seasonality effect.
This research employs the gradient descent learning (FIR.DM) approach as a learning process in a nonlinear spectral model of maximum overlapping discrete wavelet transform (MODWT) to improve volatility prediction of daily stock market prices using Saudi Arabia's stock exchange (Tadawul) data. The MODWT comprises five mathematical functions and fuzzy inference rules. The inputs are the oil price (Loil) and repo rate (Repo) according to multiple regression correlation, and the Engle and Granger Causality test Engle RF, (1987). The logarithm of the stock market price (LSCS) in Tadawul reflects the output variable. The correlation matrix reveals that there is no collinearity between the input variables, and the causality test demonstrates that the input variables significantly influence the outcome variable. According to the multiple regression, there is a substantial negative influence between Loil and LSCS but a significant positive effect between Repo and output. For the 80% dataset under ME (0.000005), MAE (0.003214), and MAPE (0.064497), the MODWT-LA8 (ARIMA(1,1,0) with drift) for the LSCS variable performs better than other WT functions. In the novel hybrid model MODWT-FIR.DM, each function's approximation coefficient (LSCS) is applied with input variables (Loil and Repo). We evaluate the performance of the proposed model (MODWT-LA8-FIR.DM) using different statistical measures (ME, RMSE, MAE, MPE) and compare it to two established models: the original FIR.DM and other MODWT-FIR.DM functions for forecasting 20% of datasets. The outcomes show that the MODWT-LA8-FIR.DM performs better than the traditional models based on lower ME (3.167586), RMSE (3.167638), MAE (3.167586), and MPE (80.860849). The proposed hybrid model may be a potential stock market forecasting model.