INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIII, Issue IX, September 2024

www.ijltemas.in Page 86

Enhancing Electricity Demand Forecasting Accuracy Through

Hybrid Models and Deep Learning Techniques: A Systematic

Literature Review

Abigail Mba Dabuoh

, Atta Yaw Agyeman

, Samuel Gbli Tetteh

Department of Computer Science and Informatics, University of Energy and Natural Resources Sunyani, Ghana

D Jarvis College of Computing and Digital Media, DePaul University, Chicago, USA

DOI: https://doi.org/10.51583/IJLTEMAS.2024.130908

Received: 16 July 2024; Accepted: 29 July 2024; Published: 05 October 2024

Abstract: This reviewed literature on electricity forecasting covers its history, terminology, and techniques. A systematic review

of existing studies highlighted key findings and future research opportunities. Conventional statistical techniques and MLA can

predict electricity demand over time with various techniques and forecasting windows tailored to data and problem specifics.

Most studies focused on STLF, often without testing techniques on MTLF and LTLF. The key findings include:

Many studies (26%) used conventional statistical methods like ARIMA, ARIMAX, and SARIMAX for electricity forecasting,

often without benchmarking algorithms. Various factors, such as time, weather, electricity price, population, and economy,

influence ELF. Weather parameters were the most commonly used predictors, though performance varied across studies. A global

increase in electricity demand has driven numerous studies, though less research has been done in low- and middle-income

countries. Deep neural networks like LSTM have been underutilised in electricity forecasting. LSTM's ability to store memory

and address the vanishing gradient problem makes it promising for future research, particularly in hybrid models combining CNN

and LSTM for forecasting peak load demand based on economic and environmental factors.

I. Introduction

The review provides a concise overview of electricity load forecasting (ELF). It covers the fundamental concepts, various types

of load forecasting, and the factors that influence electricity demand forecasting. Additionally, it reviews relevant machine

learning algorithms and related studies, concluding with a chapter summary. Electricity Demand - Global electricity consumption

has surged more rapidly than overall energy usage in recent years. From 1980 to 2013, annual global electricity consumption rose

from 7,300 TWh to 22,100 TWh. In the 21st century, the growth rate of electricity consumption has averaged 3.4% per year,

outpacing the 2.2% annual increase in overall energy consumption (Liu, 2015). According to the IEA’s Electricity Market Report,

global electricity demand, which declined by about 1% in 2020 due to the Covid-19 pandemic, was expected to grow by nearly

5% in 2021 and 4% in 2022, driven by economic recovery. Developing regions in Asia and Central and South America have seen

particularly rapid increases in electricity consumption.

In Ghana, the history of the power industry dates back to the colonial era, when energy was primarily supplied by isolated diesel

generators owned by industries, municipalities, and institutions like hospitals and schools (Kumi, 2017). Despite significant

increases in installed generation capacity from 1,730 MW in 2006 to 3,795 MW in 2016, Ghana has faced persistent energy

supply issues, resulting in an average daily economic loss of US $2.1 million. During this period, peak power consumption grew

by 50%, from 1,393 MW to 2,087 MW. The National Electrification Scheme (NES), launched in 1990, has significantly

increased electricity access, which rose from 15-20% in 1990 to 82.5% in 2016. However, Ghana may miss its goal of universal

access by 2020 by 5% unless the electrification rate increases.

Ghana's Electricity Demand and Supply Nexus - From 2006 to 2016, Ghana's peak electricity demand grew by 49.8%, from 1,393

MW to 2,087 MW, averaging an annual increase of 4.29% (Energy Commission of Ghana, 2016a; VRA, 2015; Energy

Commission of Ghana, 2017, cited by Kumi, 2017). Over the same period, generation capacity more than doubled, with an

average annual increase of 8.60%, rising from 1,730 MW to 3,759 MW. Despite this, power shortages have persisted due to

various challenges.

Ghana's gross electricity consumption fluctuated from 9,059 GWh in 2006 to 7,413 GWh in 2007 before increasing by an average

of 10.8% annually until 2014. This trend reversed with an 11.3% decline from 2014 to 2015, followed by a 15.3% rise in 2016

(Energy Commission of Ghana, 2016a; Energy Commission of Ghana, 2017, cited by Kumi, 2017).

Accurate power forecasting is crucial for effective energy resource planning and management, as it directly impacts a country's

economic activities (Hadjout et al., 2021).

Brief History of Electricity Load Forecasting (ELF): ELF involves predicting future load requirements to make informed

system expansion decisions. The concept dates back to 1965 (Heinemann and Nordmian, cited by Yang et al., 2019) and has since

evolved, yielding increasingly accurate results. ELF underpins system expansion and tariff decisions.

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIII, Issue IX, September 2024

www.ijltemas.in Page 87

Basic Concept of ELF: An electricity demand forecast model typically involves setting a learning objective, collecting and

partitioning data into training and testing sets, selecting a suitable machine learning algorithm (MLA), and evaluating the model's

performance using metrics such as RMSE, MAPE, and correlation coefficient. The predicted output is compared to the desired

output to assess the model's accuracy.

Factors Influencing ELF: Several factors influence system load behavior, necessitating a thorough understanding for accurate

forecasting. These factors include:

Weather: Parameters like temperature, humidity, and wind speed affect electrical appliance usage.

Time and Day: Load variations depend on the time of day, holidays, weekdays/weekends, and seasons.

Economic: Factors like industrialization, load management policies, and electricity pricing significantly impact load trends.

Random Disturbance: Unexpected events, such as industrial shutdowns or special events like football matches, cause load

variations.

Customer Category: Load factors vary for residential, commercial, and industrial customers, influenced by production levels,

population growth, and other demographic factors.

These factors are detailed in studies by Charytoniuk et al. (1998), Hassan et al. (2014), Mengying et al. (2019), Ruzic et al.

(2003), and Zivanovic (2002)

Table 1: Common Evaluation Metrics for Electric Load Demand Forecasters

Abbreviation

Evaluation Metric

Definition

RMSE

Root Mean Squared Error

MAE

Mean Absolute Error

MAPE

Mean Absolute Percentage Error

Nash-Sutcliffe Coefficient Radius

RPD

Relative Percentage Difference

AUC

Area Under the Curve

NMSE

Normalised Mean Squared Error

MSPE

Mean Squared Prediction Error

RMSEP

Root Mean Square Prediction Error

MBE

Mean Bias Error

Correlation Coefficient

Accuracy

Precision

PRE=TP/(TP+FP)

Recall

REC=TP/(TP+FN)

MedAE

Median Absolute Error

MedAE(y,y



)=median(

Types of ELF: Electricity demand forecasting (ELF) can be categorized by the techniques used or the forecasting duration. This

section elaborates on these categories in detail.

ELF Types Based on Forecasting Intervals (Lead Time) {ELF can be divided into very short-term, short-term, medium-term, and

long-term forecasts based on the forecasting intervals, also known as lead time. Table 2.2 summarizes the types of load

forecasting and their applications, as detailed in works by Kuster et al. (2017), Nti et al. (2019), and Nti et al. (2020).

Table 2: Summary of Load Forecasting Types

Nature of

Forecast

Lead Time

Application

Very short-term

Seconds to minutes

Generation, distribution schedules, and contingency analysis for

system security

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIII, Issue IX, September 2024

www.ijltemas.in Page 88

Short-term

Half an hour to a few

hours

Distribution of spinning reserve, operational planning, unit

commitment, maintenance schedule

Medium-term

A few days to weeks

Seasonal peak planning (winter, summer)

Long-term

Months to years

Generation growth planning

Problems in Short-Term Load Forecasting (STLF): There are several issues encountered in STLF, which are discussed below:

1. Input-Output Relationship: Most STLF methods use an artificial neural network (ANN) structure to model the input-

output relationship. However, designing this network requires detailed prior knowledge. Incorrect network design can

lead to poor predictions. Identifying significant input variables is also challenging. Too many or too few variables can

reduce accuracy. Clustering and mode recognition tools can improve results by better representing system properties,

though they still require prior knowledge for effective clustering.

2. Expert Experience: Experienced personnel in power grids and load dispatch centers often outperform computer

forecasts. Thus, expert systems and fuzzy inference systems are used, but this requires transforming expert knowledge

into a rule database.

3. Anomalous Days: Predicting unusual load days, such as holidays or extreme weather days, is difficult due to their

differing load behavior. Using historical data from the past five years can help, but load growth may still cause

dissimilarities.

4. Weather Data: Weather significantly impacts forecasting accuracy. Despite advances in weather forecasting,

inaccuracies remain. Detailed weather data is often unavailable, which can lead to errors in load forecasting.

5. Training Problems: ANN-based load forecasting involves training and predicting with two data sets: training and

testing data. Overfitting can occur if the model performs well on training data but poorly on new data. Proper training

methods are needed to avoid this.

6. Reliability: Economic development often outpaces power investment, leading to energy shortages. Demand-side

management can disrupt natural load curves, complicating forecasting. Ensuring reliable data and removing noise is

essential for accurate forecasts.

Requirements of STLF: A sound STLF system should meet the following requirements (Dewari & Bhandari, 2015):

1. Accuracy: The primary requirement for STLF is accuracy, as it underpins economic dispatch, system reliability, and

electricity market trading.

2. Speed: The STLF program should utilize the latest historical and weather data to increase accuracy and reduce

computation time. Programs with long training times should be replaced with faster techniques that maintain accuracy.

3. Detection of Bad Data: Modern STLF systems should automatically detect and eliminate erroneous data, reducing the

burden on operators.

4. User-Friendliness: The load forecasting interface should be intuitive, allowing users to easily define forecasts and view

results both numerically and graphically.

5. Automatic Forecasting: To reduce the risk of inaccurate forecasts, STLF systems should automatically generate final

results based on past performance, without requiring operator intervention.

ELF Based on Techniques: ELF techniques can be broadly classified into three categories: correlation, extrapolation, and a

combination of both (Eeeguide.com, 2014; Nti et al., 2019).

 Correlation Techniques: These relate system load to various economic and demographic factors, helping forecasters

understand the relationship between load growth and measurable factors.

 Extrapolation Techniques: These include time-series or conventional methods that fit trend curves to historical data. The

forecast is obtained by evaluating the trend curve function at the desired future point.

No single forecasting method is universally applicable. ELF techniques are categorized into data-driven (AI) techniques and

engineering techniques, though there is no consensus on which is superior.

Machine Learning Algorithms: Machine learning (ML), a subset of artificial intelligence (AI), provides methods for solving

complex problems (Stanisavljevic & Spitzer, 2016). ML enables software and machines to learn from experience without explicit

programming. According to literature (Nti et al., 2019; Simeone, 2018; Stanisavljevic & Spitzer, 2016), ML algorithms can be

categorized into:

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIII, Issue IX, September 2024

www.ijltemas.in Page 89

 Supervised Learning (SL): The algorithm learns from labeled data (training data) to build a model that predicts outcomes

for new data.

 Unsupervised Learning (UL): The algorithm learns from unlabeled data to identify patterns and build models.

 Semi-Supervised Learning (SSL): Combines aspects of supervised and unsupervised learning.

 Reinforcement Learning (RL): The algorithm interacts with an environment to achieve a goal, learning from actions

without a teacher.

Examples of ML algorithms include General Regression Neural Network (GRNN), Support Vector Machine (SVM), Artificial

Neural Networks (ANN), and many others (Nti et al., 2021). Figure 2.2 illustrates ML algorithm classifications and examples.

A Systematic Review of Related Works:

This section provides a systematic review of studies on ELF. According to Keele (2007), a systematic literature review (SLR)

aims to identify, assess, and discuss relevant works to answer research questions. An SLR must be comprehensive and unbiased

to be scientifically valuable.

Kamilaris and Prenafeta-Boldú (2018) stated that SLR helps develop essential insights and identify potential research gaps. This

study adopts a 3-stage SLR process (Ardabili et al., 2020; Mosavi et al., 2019; Sharma et al., 2020)

1. Pre-Operational (Review Planning):

2. Operational (Conducting the Review):

3. Post-Operational (Review Findings):

The aim is to answer questions such as: What factors influence ELF? What ML algorithms are used for electricity demand

forecasting? What are the strengths and weaknesses of these algorithms? What timeframes and evaluation metrics are used in

ELF?

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIII, Issue IX, September 2024

www.ijltemas.in Page 90

II. Review Outcome

This section outlines the results of the SLR, highlighting key findings and gaps in the literature that justify the significance of this

study.

Year-wise and Publisher-wise Distribution of Articles

Recently, electricity demand forecasting has gained significant attention in both academia and industry. Table 6 illustrates the

yearly distribution of articles, showing a growing interest in electricity load forecasting over the years.

Figure 1 shows the distribution of articles by publisher, indicating that publishers like IEEE, Elsevier, and Springer recognize the

importance of studies in this field. IEEE had the highest number of publications, likely due to its focus on engineering works.

Analysis Based on Used Forecaster: presents the most commonly used forecasting techniques in electricity demand literature.

Seasonal ARIMA and ANN are the most popular classical methods for electricity load forecasting. The ARIMA model is favored

for its ability to handle seasonal components in LTLF, where variations are less frequent. ANN and SVM are prominent

computational intelligence techniques effective in modeling the non-linearity and complex relationships in electricity demand

influenced by economic and environmental factors. However, existing AI-based models mainly address VSTLF challenges, while

traditional statistical methods are static and rely on historical data (Bedi & Toshniwal, 2019).

Combining multiple forecasting techniques into hybrid models is believed to enhance accuracy. Hybrid models integrate various

ML algorithms to leverage their strengths. Studies like Ganguly et al. (2020), Haq & Ni (2019), Jarndal & Hamdan (2017), Rusli

et al. (2019), and Yildiz et al. (2017) demonstrate that hybrid techniques outperform individual models in electricity forecasting.

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIII, Issue IX, September 2024

www.ijltemas.in Page 91

Figure 2: Most Used Forecasters in Electricity Demand Forecasting: Deep neural networks, such as ANN (including BPNN and

CNN), have been less frequently used. Only a few studies have applied DBN (Dedinec et al., 2016; Haq & Ni, 2019) and LSTM

(Bedi & Toshniwal, 2019). PCA is widely used for dimensionality reduction (Aneiros et al., 2016; De Felice et al., 2015; Yildiz et

al., 2017), while ensemble learning techniques, believed to be superior to single learners, have also been adopted (de Oliveira &

Cyrino Oliveira, 2018; Divina et al., 2018; Pannakkong et al., 2018).

Despite the popularity of ARIMA and ANN, Pereira et al. (2015) found that the Fuzzy Inference System (FIS) performed better

than SARIMAX. Fu et al. (2015) reported that SVM outperformed ARIMAX, DT, and ANN. Conversely, Azad et al. (2018)

found that ANN optimized by the sine-cosine algorithm and GH outperformed SVM. These mixed findings suggest that the

choice of forecasting method depends on factors like data characteristics, task type (regression or classification), and forecasting

period.

Figure 3 highlights the evaluation metrics commonly used to validate forecasting performance in literature, with MAPE, RMSE,

and MAE being the most frequently used due to their suitability for regression analysis.

Figure 2.9: Commonly Used Evaluation Metrics

III. Summary of Analyzed Articles

Table 2.9 summarizes the analyzed articles. A significant proportion (56%) used correction techniques, 41% used extrapolation

techniques, and 3% combined both. The forecasting windows showed 55% used STLF, 18% VSTLF, 16% LTLF, and 11%

MTLF.

Of the 37 articles, 82% aimed to forecast future electricity demand using regression analysis, with few studies (5%) adopting

classification techniques, and 13% using clustering methods. The origin of the analyzed articles showed 15% from Australia, 12%

from India, and 7% each from Thailand, China, and Italy.

Figure 2.7 summarizes the most common predictors for electricity demand. Various factors like time, day, temperature, weather,

and economic conditions influence ELF. Weather parameters, especially temperature, were the most frequently used predictors

for STLF, MTLF, and LTLF. Studies often used historical load demand and weather variables (temperature, humidity,

precipitation, and solar gain) as predictors. Economic variables and historical load demand were also commonly used, reflecting

the interconnected nature of electricity consumption with human activities, population, and economic status.

The significance of diverse forecasting methodologies and their applications has been highlighted by this systematic assessment

of the literature, which delves into the changing landscape of energy demand forecasting. Many machine learning techniques,

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIII, Issue IX, September 2024

www.ijltemas.in Page 92

including ANN and SVM, as well as conventional statistical techniques like ARIMA, have been applied extensively and

demonstrated significant forecasting performance. However, these traditional approaches are unable to completely address the

continuous issues posed by the complexity and non-linearity of electricity consumption patterns.

According to the assessment, there is a growing interest in hybrid models, which mix several algorithms to take use of each one's

unique characteristics and increase predicting accuracy. Research employing hybrid methodologies has shown to perform better

than single-method approaches, indicating a promising avenue for further investigation. The article also emphasizes the

unrealized promise of deep learning methods, especially LSTM networks, which provide sophisticated capabilities in processing

sequential data and addressing problems such as the vanishing gradient problem.

The examined research identify crucial predictors like meteorological characteristics, economic indicators, and historical load

data, which are important elements determining electricity demand. The emphasis on obtaining accurate and trustworthy forecasts

is reflected in the widespread use of regression analysis and error metrics like MAPE, RMSE, and MAE for performance

validation.

The research identifies shortcomings in the application of contemporary forecasting approaches despite notable developments,

particularly in low- and middle-income nations. This emphasizes how more study is required to create reliable models that can be

applied in a variety of geographic and economic circumstances.

Conclusion, there is a great deal of potential for improving the precision and dependability of electricity demand projections

through the incorporation of deep learning methods and hybrid models. Subsequent investigations have to center on delving into

these sophisticated techniques, refining their implementation throughout various projection periods, and attending to the

particular requirements of marginalized areas. By doing this, the industry can get one step closer to creating all-inclusive

solutions that facilitate effective global energy planning and management.

References

1. Aneiros, G., Vilar, J., & Raña, P. (2016). Short-term forecast of daily curves of electricity demand and price.

International Journal of Electrical Power & Energy Systems, 80, 96–108. https://doi.org/10.1016/j.ijepes.2016.01.034

2. Ardabili, S., Mosavi, A., & Várkonyi-Kóczy, A. R. (2020). Systematic Review of Deep Learning and Machine Learning

Models in Biofuels Research. In Melting Threshold and Thermal Conductivity of CdTe Under Pulsed Laser Irradiation

(Vol. 101, pp. 29–42). Springer. https://doi.org/10.1007/978-3-030-36841-8_10

3. Azad, M. K., Uddin, S., & Takruri, M. (2018). Support vector regression based electricity peak load forecasting. 2018

11th International Symposium on Mechatronics and Its Applications (ISMA), 2018-Janua, 1–5.

https://doi.org/10.1109/ISMA.2018.8330143

4. Bedi, J., & Toshniwal, D. (2019). Deep learning framework to forecast electricity demand. Applied Energy, 238(October

2018), 1312–1326. https://doi.org/10.1016/j.apenergy.2019.01.113

5. Charytoniuk, W., Chen, M. S., & Van Olinda, P. (1998). Nonparametric regression based short-term load forecasting.

IEEE Transactions on Power Systems, 13(3), 725–730. https://doi.org/10.1109/59.708572

6. Dedinec, A., Filiposka, S., Dedinec, A., & Kocarev, L. (2016). Deep belief network based electricity load forecasting:

An analysis of Macedonian case. Energy, 115, 1688–1700. https://doi.org/10.1016/j.energy.2016.07.090

7. De Felice, M., Alessandri, A., & Catalano, F. (2015). Seasonal climate forecasts for medium-term electricity demand

forecasting. Applied Energy, 137, 435–444. https://doi.org/10.1016/j.apenergy.2014.10.030

8. de Oliveira, E. M., & Cyrino Oliveira, F. L. (2018). Forecasting mid-long term electric energy consumption through

bagging ARIMA and exponential smoothing methods. Energy, 144, 776–788.

https://doi.org/10.1016/j.energy.2017.12.049

9. Dewari, S. S., & Bhandari, V. (2015). Electric load forecasting based on locally weighted support vector regression.

International Journal for Scientific Research & Development, 40(4), 2321–0613.

10. Divina, F., Gilson, A., Goméz-Vela, F., García Torres, M., & Torres, J. (2018). Stacking Ensemble Learning for Short-

Term Electricity Consumption Forecasting. Energies, 11(4), 949. https://doi.org/10.3390/en11040949

11. Eeeguide.com. (2014). Forecasting Methodology. http://www.eeeguide.com/forecasting-methodology/

12. Fu, Y., Li, Z., Zhang, H., & Xu, P. (2015). Using Support Vector Machine to Predict Next Day Electricity Load of

Public Buildings with Sub-metering Devices. Procedia Engineering, 121, 1016–1022.

https://doi.org/10.1016/j.proeng.2015.09.097

13. Ganguly, A., Goswami, K., & Kumar Sil, A. (2020). WANN and ANN based Urban Load Forecasting for Peak Load

Management. 2020 IEEE Calcutta Conference (CALCON), 402–406.

https://doi.org/10.1109/CALCON49167.2020.9106520

14. Hadjout, D., Torres, J. F., Troncoso, A., Sebaa, A., & Martínez-Álvarez, F. (2021). Electricity consumption forecasting

based on ensemble deep learning with application to the algerian market. Energy, 123060.

https://doi.org/10.1016/j.energy.2021.123060

15. Haq, M. R., & Ni, Z. (2019). A New Hybrid Model for Short-Term Electricity Load Forecasting. IEEE Access, 7,

125413–125423. https://doi.org/10.1109/ACCESS.2019.2937222

16. Hassan, S., Khosravi, A., Jaafar, J., & Raza, M. Q. (2014). Electricity load and price forecasting with influential factors

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIII, Issue IX, September 2024

www.ijltemas.in Page 93

in a deregulated power industry. 2014 9th International Conference on System of Systems Engineering (SOSE), 79–84.

https://doi.org/10.1109/SYSOSE.2014.6892467

17. Jarndal, A., & Hamdan, S. (2017). Forecasting of peak electricity demand using ANNGA and ANN-PSO approaches.

2017 7th International Conference on Modeling, Simulation, and Applied Optimization (ICMSAO), 1–5.

https://doi.org/10.1109/ICMSAO.2017.7934842

18. Kamilaris, A., & Prenafeta-Boldú, F. X. (2018). Deep learning in agriculture: A survey. Computers and Electronics in

Agriculture, 147(July 2017), 70–90. https://doi.org/10.1016/j.compag.2018.02.016

19. Keele, S. (2007). Guidelines for performing Systematic Literature Reviews in Software Engineering. In EBSE Technical

Report EBSE-2007-01: Vol. 2.3 (Issue 5).

20. Kumi, E. N. (2017). The Electricity Situation in Ghana: Challenges and Opportunities. Center for Global Development,

September. www.cgdev.org

21. Kuster, C., Rezgui, Y., & Mourshed, M. (2017). Electrical load forecasting models: A critical systematic review.

Sustainable Cities and Society, 35, 257–270. https://doi.org/10.1016/j.scs.2017.08.009

22. Liu, Z. (2015). Global Energy Development: The Reality and Challenges. In Global Energy Interconnection (pp. 1–64).

Elsevier. https://doi.org/10.1016/B978-0-12-804405-6.00001-4

23. Mengying, H., Jiandong, D., Zequan, H., Peng, W., Shuai, F., Peijia, H., & Chaoyuan, F. (2019). Monthly Electricity

Forecast Based on Electricity Consumption Characteristics Analysis and Multiple Effect Factors. 2019 IEEE 8th

International Conference on Advanced Power System Automation and Protection (APAP), 1814–1818.

https://doi.org/10.1109/APAP47170.2019.9224784

24. Mosavi, A., Salimi, M., Faizollahzadeh Ardabili, S., Rabczuk, T., Shamshirband, S., & Varkonyi-Koczy, A. (2019).

State of the Art of Machine Learning Models in Energy Systems, a Systematic Review. Energies, 12(7), 1301.

https://doi.org/10.3390/en12071301

25. Nti, I. K., Adekoya, A. F., & Weyori, B. A. (2019a). A systematic review of fundamental and technical analysis of stock

market predictions. Artificial Intelligence Review, 53(4), 3007–3057. https://doi.org/10.1007/s10462-019-09754-z

26. Nti, I. K., Adekoya, A. F., & Weyori, B. A. (2019b). Random Forest Based Feature Selection of Macroeconomic

Variables for Stock Market Prediction. American Journal of Applied Sciences, 16(7), 200–212.

https://doi.org/10.3844/ajassp.2019.200.212

27. Nti, I. K., Adekoya, A. F., & Weyori, B. A. (2021). A novel multi-source information-fusion predictive framework based

on deep neural networks for accuracy enhancement in stock market prediction. Journal of Big Data, 8(1), 17.

https://doi.org/10.1186/s40537-020-00400-y

28. Pannakkong, W., Sriboonchitta, S., & Huynh, V.-N. (2018). An Ensemble Model of Arima and Ann with Restricted

Boltzmann Machine Based on Decomposition of Discrete Wavelet Transform for Time Series Forecasting. Journal of

Systems Science and Systems Engineering, 27(5), 690–708. https://doi.org/10.1007/s11518-018-5390-8

29. Pereira, C. M., Almeida, N. N. de, & Velloso, M. L. F. (2015). Fuzzy Modeling to Forecast an Electric Load Time

Series. Procedia Computer Science, 55(Itqm), 395–404. https://doi.org/10.1016/j.procs.2015.07.089

30. Rusli, R., Hidayanto, A. N., & Ruldeviyani, Y. (2019). Consumption Prediction on Steam Power Plant Using Data

Mining Hybrid Particle Swarm Optimization (PSO) and Auto Regressive Integrated Moving Average (ARIMA). 2019

International Workshop on Big Data and Information Security (IWBIS), 15–20.

https://doi.org/10.1109/IWBIS.2019.8935844

31. Ruzic, S., Vuckovic, A., & Nikolic, N. (2003). Weather sensitive method for short term load forecasting in electric

power utility of serbia. IEEE Transactions on Power Systems, 18(4), 1581–1586.

https://doi.org/10.1109/TPWRS.2003.811172

32. Sharma, R., Kamble, S. S., Gunasekaran, A., Kumar, V., & Kumar, A. (2020). A systematic literature review on

machine learning applications for sustainable agriculture supply chain performance. Computers and Operations

Research, 119, 104926. https://doi.org/10.1016/j.cor.2020.104926

33. Simeone, O. (2018). A Very Brief Introduction to Machine Learning with Applications to Communication Systems.

IEEE Transactions on Cognitive Communications and Networking, 4(4), 648–664.

https://doi.org/10.1109/TCCN.2018.2881442

34. Stanisavljevic, D., & Spitzer, M. (2016). A Review of Related Work on Machine Learning in Semiconductor

Manufacturing and Assembly Lines. August 2018.

35. Yang, A., Li, W., & Yang, X. (2019). Short-term electricity load forecasting based on feature selection and Least

Squares Support Vector Machines. Knowledge-Based Systems, 163, 159–173.

https://doi.org/10.1016/j.knosys.2018.08.027

36. Yildiz, B., Bilbao, J. I., & Sproul, A. B. (2017). A review and analysis of regression and machine learning models on

commercial building electricity load forecasting. Renewable and Sustainable Energy Reviews, 73(December 2016),

1104–1122. https://doi.org/10.1016/j.rser.2017.02.023

37. Zivanovic, R. (2002). Nonparametric trend model for short term electricity demand forecasting. Fifth International

Conference on Power System Management and Control, 2002, 347–352. https://doi.org/10.1049/cp:20020060