FORECASTING

Forecasting is the process of making predictions of the future based on past and present data and most
commonly by analysis of trends. A commonplace example might be estimation of some variable of
interest at some specified future date. Prediction is a similar, but more general term. Both might refer to
formal statistical methods employing time series, cross-sectional or longitudinal data, or alternatively to
less formal judgmental methods. Usage can differ between areas of application: for example,
in hydrology the terms "forecast" and "forecasting" are sometimes reserved for estimates of values at
certain specific future times, while the term "prediction" is used for more general estimates, such as the
number of times floods will occur over a long period.

CLASSIFICATION OF FORECASTING TECHNIQUES

Qualitative vs. quantitative methods
Qualitative forecasting techniques are subjective, based on the opinion and judgment of consumers,
experts; they are appropriate when past data are not available. They are usually applied to intermediate- or
long-range decisions. Examples Delphi method, market research, historical lifecycle analogy.
Quantitative forecasting models are used to forecast future data as a function of past data. They are
appropriate to use when past numerical data is available and when it is reasonable to assume that some of
the patterns in the data are expected to continue into the future. These methods are usually applied to
short- or intermediate-range decisions. Examples of quantitative forecasting methods are last period
demand, simple and weighted N-Period moving averages, simple exponential smoothing, poisson process
model based forecasting [2] and multiplicative seasonal indexes. Previous research shows that different
methods may lead to different level of forecasting accuarcy. For example, GMDH neural network was
found to have better forecasting performance than the classical forecasting algorithms such as Single
Exponential Smooth, Double Exponential Smooth, ARIMA and back-propagation neural network.
Average approach
In this approach, the predictions of all future values are equal to the mean of the past data. This approach can
be used with any sort of data where past data is available. In time series notation:

Although the time series notation has been used here, the average approach can also be used for cross-sectional
data (when we are predicting unobserved values; values that are not included in the data set). Then, the
prediction for unobserved values is the average of the observed values.

Naïve approach
Naïve forecasts are the most cost-effective forecasting model, and provide a benchmark against which
more sophisticated models can be compared. This forecasting method is only suitable for time
series data Using the naïve approach, forecasts are produced that are equal to the last observed value. This
method works quite well for economic and financial time series, which often have patterns that are
difficult to reliably and accurately predict.[4] If the time series is believed to have seasonality, seasonal
naïve approach may be more appropriate where the forecasts are equal to the value from last season. The
naïve method may also use a drift, which will take the last observation plus the average change from the
first observation to the last observation.[4] In time series notation:

Drift method
A variation on the naïve method is to allow the forecasts to increase or decrease over time, where the
amount of change over time (called the drift) is set to be the average change seen in the historical data. So
the forecast for time T+ h is given by

This is equivalent to drawing a line between the first and last observation, and extrapolating it into the
future.
Time series methods
Time series methods use historical data as the basis of estimating future outcomes.
Causal / econometric forecasting methods
Some forecasting methods try to identify the underlying factors that might influence the variable that is
being forecast. For example, including information about climate patterns might improve the ability of a
model to predict umbrella sales. Forecasting models often take account of regular seasonal variations. In
addition to climate, such variations can also be due to holidays and customs: for example, one might
predict that sales of college football apparel will be higher during the football season than during the off
season.
Several informal methods used in causal forecasting do not employ strict algorithms , but instead use the
judgment of the forecaster. Some forecasts take account of past relationships between variables: if one
variable has, for example, been approximately linearly related to another for a long period of time, it may
be appropriate to extrapolate such a relationship into the future, without necessarily understanding the
reasons for the relationship.
Judgmental methods
Judgmental forecasting methods incorporate intuitive judgement, opinions and
subjective probability estimates. Judgmental forecasting is used in cases where there is lack of historical
data or during completely new and unique market conditions.
Artificial intelligence methods
 Artificial neural networks
 Group method of data handling
 Support vector machines
Often these are done today by specialized programs loosely labeled
 Data mining
 Machine Learning
 Pattern Recognition

CLASSIFICATION OF LOAD FORECASTING TECHNIQUES

Electricity demand forecasts are extremely important for energy suppliers and other participants in
electric energy generation, transmission, distribution and markets. Accurate models for electric power
load forecasting are essential to the operation and planning of a utility company. Load forecasts are
extremely important for energy suppliers and other participants in electric energy generation,
transmission, distribution and markets. This paper presents a review of electricity demand forecasting
techniques. The various types of methodologies and models are included in the literature. Load
forecasting can be broadly divided into three categories: short-term forecasts which are usually from one
hour to one week, medium forecasts which are usually from a week to a year, and long-term forecasts
which are longer than a year. Based on the various types of studies presented in these papers, the load
forecasting techniques may be presented in three major groups: Traditional Forecasting technique,
Modified Traditional Technique and Soft Computing Technique.

 Adaptive Demand Forecasting

Demand forecasting model parameters are automatically corrected to keep track of the changing load
conditions. Hence Demand forecasting is adaptive in nature and can also be used as an on-line software
package in the utilities control system. Next state vector is estimated using current prediction error and
the current weather data acquisition programs. State vector is determined by total historical data set
analysis. Switching between multiple and adaptive regression analysis is possible in this mode. The same
model as in the multiple regression section, by equation given below is used in this model .

-vector of adapted variables such as time, temperature, light intensity, wind speed, humidity, day type
(workday, weekend), etc., at -transposed vector of regression coefficients and et-Model error at time t.
Lu developed an adaptive Hammerstein model with an orthogonal escalator structure as well as a lattice
structure for joint processes. This model used a joint Hammerstein nonlinear time-varying functional
relationship between load and temperature. This algorithm performed better than the commonly used
RLS (Recursive Least-square) algorithm. Grady enhanced and applied the algorithm developed by Lu.
An improvement was obtained in the ability to forecast total system hourly load as far as 5 days
McDonald , presented an adaptive-time series model and simulated the effects of a direct load control
strategy. A composite model for load prediction composed of three components (nominal load, type load
and residual load) was developed by Park in .

 Stochastic Time Series

The Time series methods appear to be among the most popular approaches that applied to STLF. Time
series methods are based on the assumption that the data have an internal structure, such as
autocorrelation, trend or seasonal variation. The first impetus of the approach is to accurately assemble a
pattern matching available data and then obtain the forecasted value with respect to time using the
established model. The next subsection discusses some of the time series models used for load
forecasting.

 Autoregressive (AR) Model

Auto-Regressive (AR) model can be used to model the load profile, If the load is assumed to be a linear
combination of previous loads, which is given by Liu as:

Where, Lk is the predicted load at time k (min), is a random load disturbance, αi , i =1…m are unknown
coefficients and above given equation is the auto regressive model of order m. The unknown coefficients
in equation can be tuned on-line using the well-known least mean square (LMS) algorithm of Mbamalu
and El-Hawary . Huang and Zhao proposed an autoregressive model with an optimum threshold
stratification algorithm and two periodical autoregressive (PAR) models for hourly load forecasting
respectively

 Support Vector Machine based Techniques

Vapnik was the first to introduce SVM; it is a novel powerful machine learning method based on
statistical learning theory (SLT), which analyzes data and recognizes patterns, used for classification and
regression analysis. They combine generalization control with a technique to address the curse of
dimensionality . B.J.Chen et.al proved that temperature and other climate information is not much useful
for mid-term load forecasting and introduction of time series forecasting may improve the results . F. E.
H. Tay and L. J. Cao modified risk function of conventional support vector machines by penalizing
insensitive errors more heavily than the distant insensitive errors, they named this method as C-ascending
support vector machine. They conclude by a test that the C-ascending support vector machines with the
actually ordered sample data consistently forecast better than the standard support vector machines. X.
Tao et.al proposed a SVM based strategy to rand individual components according to their influence on
the load forecasting by limiting the number of features that cuts down the model capacity . To estimate
the relations between input and output variables Lee & Song further modified the Support Vector
Machine (SVM) by using an empirical inference model. This method was derived by modifying the risk
function of the standard SVM by using the concept of Locally Weighted Regression. The proposed
method proves useful to be in the field of process monitoring, optimization and quality control . G. S. Hu
et.al presented a new short term load forecasting method by conjunctive use of fuzzy C-mean clustering
algorithm and weighted support vector machines (WSVMs). They clustered input samples according to
the similarity degree [35]. Ying-Chun Guo showed that SVM based model provides a promising
arithmetic to forecasting electricity load than artificial neural network. The model overcomes the
disadvantages of general artificial neural network (ANN), such as it is not easy to converge, liable to trap
in partial minimum and unable to optimize globally, and the generalization of the model is not good, etc .
Jingmin Wang et.al proposed a new optimal model which is based on Stimulated Annealing Particle
Swarm Optimization Algorithm (SAPSO) that combines the advantages of PSO and SA algorithm. The
new algorithm is employed to choose the parameters of a SVM model. The model is proved to be able to
enhance the accuracy and improved the convergence ability and reduced operation time by numerical
experiment. Jingmin Wang et al., presented A short-term load forecasting model based on SVM with
Adaptive Quantum-behaved Particle Swarm Optimization Algorithm (AQPSO) .

 Soft Computing Techniques

It is a fact that every system is pervasively imprecise, uncertain and hard to be modelled precisely. A
flexible approach called Soft Computing technique has emerged to deal such models effectively and most
efficiently on research scenario. It has been very widely in use over the last few decades. Soft computing
is an emerging approach which parallels the remarkable ability of the human mind to reason and learn in
an environment of uncertainty and imprecision. It is fast emerging as a tool to help computer-based
intelligent systems mimic the ability of the human mind to employ modes of reasoning that are
approximate rather than exact. The basic theme of soft computing is that precision and certainty carry a
cost and that intelligent systems should exploit, wherever possible, the tolerance for imprecision and
uncertainty. Soft computing constitutes a collection of disciplines which include fuzzy logic (FL), neural
networks (NNs), evolutionary algorithms (EAs) like genetic algorithms (GAs) etc. Natural intelligence is
the product of millions of years of biological evolution. Simulating complex biological evolutionary
processes may lead us to discover, how evolution propels living systems toward higher-level of
intelligence. One of the newer and relatively simple optimization approaches is the GA which is based on
the evolutionary principle of natural selection. Perhaps one of the most attractive qualities of GA is that it
is a derivative free optimization tool. The demand/ load forecasting techniques are also developed based
on the following soft.

 Genetic Algorithms (Long-term Forecasts)

The other artificial intelligence-based technique involves the use of genetic algorithms, and it deals with
the utilization of genetic programming in predicting long-term consumption of electricity. The results
obtained from this procedure are reliable as the method, if successfully undertaken, usually has a low
chance of errors. This approach also uses a genetic neural network model that can adapt and learn, which
means that it is robust. At this point, the model includes the ideas and experiences of experts in the field
in order to create a comprehensive effect whereby, all variables that affect power load are reflected. This
examination makes it apparent that the technique uses a numerical optimization approach. The main
advantage of using genetic algorithms is that they include all aspects of computing such as imprecision,
non-linearity, robustness, as well as uncertainty.

 Support Vector Machine(SVM)

According to Ghods (254), Support Vector Machine is a method used for classifying information or data,
to obtain good generalization on a limited number of learning patterns. It is a technique based on a
concept known as structural risk minimization. For instance, recurrent neural network is part
of this technique and runs under the idea that every unit is an output of the entire network. There are two
main categories for SMVs namely Support Vector Classification (SVC) and Support Vector
Regression(SVR). SVM is a learning system that applies a high dimensional feature space and yields
prediction or forecasting functions that are expanded on a subset of support vectors. SVM can generalize
complicated gray level structures with only a very few support vectors and thus provides a new
mechanism for image compression as pointed out by Basak et al. (2007).
 Exponential smoothing
Exponential smoothing is one of the approaches used for load forecasting. In this method, first load is
model based on previous data, then to use this model to predict the future load. In Moghram and
Rahman’s exponential smoothing model, the load at time t, y(t), is modelled using a fitting function and is
expressed in the form [9]:

Where, f(t)Fitting function vector of the process, β(t)-Coefficient of vector, n(t)White noise and T-
Transpose operator. The Winter’s method is one of existing exponential smoothing methods having
capacity to analyze seasonal time series directly. It is based on three smoothing constants for stationary,
trend and seasonality. Barakat [14] analyzed the result of the model and conclude that unique pattern of
energy and demand pertaining to fast growing areas was difficult to analyze and predict by direct
application of the Winter’s method. Exponential smoothing was augmented with power spectrum analysis
and adaptive autoregressive modelling in El-Keib hybrid approach. Infield and Hill optimal smoothing
based trend removal technique has been shown to compare favorably with conventional methods of load
forecasting.

 Fuzzy Logic

The fuzzy logic model is a rule-based system that applies fuzzy rules, which are control decision
mechanisms. The rules usually adjust the impacts that definite stimuli have on energy demand as a factor.
The model maps on a set of input variables to a set of output variable simply using the IF-THEN logic
statement to predict and forecasting. These types of mappings allow the incorporation of expert
knowledge with Fuzzy Logic models. These models can provide algorithms that transform linguistic
strategy obtained from experts into automatic strategies. Fuzzy rules are often integrated with neural
networks so as to train artificial neural networks, as well as acquire accurate results in terms of load
demand forecasting. The advantage of this hybrid system is that it utilizes the benefits of both artificial
neural networks and fuzzy inferences thereby, formalizing and handling knowledge and experience of
expert forecasters (Alfares and Nazeeruddin, 2007).

 Expert Systems

Expert systems are known to apply rules which have been generated by experts in the field. These
procedures and rules are then transformed into software, which can automatically make forecasts about
electricity demand. However, in order for the software to be efficient, expert forecasters will have to work
hand in hand with software developers to include all expert information into the software. Here, software
developers will be involved in coding all the information obtained from experts in forming software rules.
The advantage of utilizing this method is that it is fast, accurate and easy to use when forecasting
information about electricity demand (Ghods, 2011). This is because it involves the use of software
applications, which usually run at the click of a button. F. The Similar Day Approach This technique
involves searching for historical information, which has similar traits with the day forecasted. These
characteristics may include the weather, the date and even month. In this case, load of a day, which has
got similar characteristics, will be considered to be the forecast. The technique can as well apply linear
techniques whereby, the forecast can be obtained from a group of days with similar traits(Islam, 2011).

 The Time Series Method

This technique is based on the concept that information obtained on electricity demand usually has an
internal structure, which might be in the form of an autocorrelation, seasonal variation or trend. These
methods explore and detect the internal structure and are utilized in such fields as digital signaling,
processing, electric load forecasting, as well as economics. For instance, Autoregressive Moving
Average(ARMA), Autoregressive Integrated Moving Average (ARIMA) and Autoregressive Moving
Average with Exogenous Variables (ARMA-EOV) are examples of time series techniques.

 The Econometric Approach

One method used in forecasting, which applies statistical techniques is known as the econometric
approach. This method applies a variety of techniques such as time series, regression, statistical learning
algorithms and fuzzy logic. This approach integrates statistical methods and economic theory when
forecasting demand for electricity. Additionally, this method attempts to obtain estimates observed
between energy consumption, which is the independent variable, and all other factors that affect
consumption, which represent the dependent variables.

 Iterative Reweighted Least-Squares

Mbamalu and El-Hawary used iteratively reweighted least-squares procedure to identify the
model order and parameters. The method uses an operator that controls one variable at a time and
determines optimal starting point. Autocorrelation function and the partial autocorrelation
function of the resulting differenced past load data is utilized to identify a suboptimal model of
the load dynamics. A three-way decision variable is formed by the weighting function, the tuning
constants and the weighted sum of the squared residuals in identifying an optimal model and the
subsequent parameter estimates. Consider the parameter estimation problem involving the linear
measurement equation:

Where, Y is an n x 1 vector of observations, X is an n x p matrix of known coefficients (based on

previous load data), β is a p x 1 vector of the unknown parameters and e is an n x 1 vector of random
errors. Results are more accurate when the errors are not Gaussian. Iterative methods are used to find β.
Newton method /alternatively Beaton-Turkey iterative reweighted least-square’s algorithm (IRLS) can be
applied if β is known. Mbamalu, El- Hawary [8] enhanced this work by introducing an interactive
approach employing least-squares and the IRLS procedure for estimating the parameters of a seasonal
multiplicative autoregressive model.

Conclusions
Load forecasting or prediction is a useful mechanism as it enables electric power companies to save in
terms of cost and efficiency. Most companies utilize load forecasting techniques based on historical data
on consumers by analyzing their electricity utilization of application. Some techniques are complex
because the factors utilized as the load frequently have complex variables. This implies that the more
complex the method applied, the more accurate the results obtained. This automatically shows that
artificial-based techniques will produce the most accurate results as they incorporate all aspects of factors
such as the social, environment and the economy. These techniques are observed to employ complex
techniques, which identify both linear and non-linear relationships that can be seen between electricity
consumption, as well as factors affecting consumption. Furthermore, these techniques are mostly used by
experts as they usually rely on their experience and opinions in order to be accurate. In the long run,
experts in this field will have to analyze these variables at length in order to have a clear understanding as
to how they behave, in an effort to create advanced methods of forecasting load or electricity demand.

REFERENCES

[1] G. A. N. Mbamalu and M. E. El-Hawary, “Load forecasting viasuboptimal seasonal autoregressive

models and iteratively reweightedleast squares estimation,” IEEE Trans. Power Systems, vol. 8, no. 1,
pp.343–348, 1993.
[2]D. G. Infield and D. C. Hill, “Optimal smoothing for trend removal in short term electricity demand
forecasting,” IEEE Trans. Power Systems,vol. 13, no. 3, pp. 1115–1120, 1998.

[3] G. M. Jenkins, “Practical experiences with modeling and forecast,” Time Series., 1979.

[4] T. Haida and S. Muto, “Regression based peak load forecasting using a transformation technique,”
IEEE Trans. Power Systems, vol. 9, no. 4, pp. 1788–1794, 1994.

[5] A. Harvey and S. J. Koopman, “Forecasting hourly electricity demand using time-varying splines,” J.
American Stat. Assoc., vol. 88, no. 424, pp. 1228–1236, 1993.
[6] H. T. Yang and C. M. Huang, “A new short-term load forecasting approach using self-organizing
fuzzy ARMAX models,” IEEE Trans. Power Systems, vol. 13, no. 1, pp. 217–225, 1998.
[7] H. T. Yang, C. M. Huang, and C. L. Huang, “Identification of ARMAX model for short term load
forecasting: An evolutionary programming approach,” IEEE Trans. Power Systems, vol. 11, no. 1, pp.
403–408, 1996.
[8] G. Gross G. Gross, and F. D. Galiana, “Short Term Load Forecasting. Proceedings of the IEEE," Vol.
75, pp. 1558-1573, 1987.
[9] Qia Ding, "Long-Term Load Forecast Using Decision Tree Method," Power Systems Conference and
Exposition, PSCE 06, IEEE PES, Vol.1, pp.1541-1543, 2006.

[10] G. T. Heineman, D. A. Nordman and E. C. Plant, "The Relationship Between Summer Weather and
Summer Loads - A Regression Analysis," IEEE Transaction Power Apparatus System, Vol. PAS-85,
No.11, pp.1144-1154, 1966.

[11] J. H. Park, Y. M. Park and K. Y. Lee, “Composite Modeling for Adaptive Short-Term Load
Forecasting," IEEE Transaction Power System, vol.6, no.2, pp.450–457, May 1991.

[12] G. A. N. Mbamalu and M. E. El-Hawary, “Load Forecasting Via Suboptimal Seasonal

Autoregressive Models And Iteratively Reweighted Least Squares Estimation," IEEE Transaction on
Power System, Vol.8, pp.343-348, 1992.

[13] I. Moghram, and S. Rahman, “Analysis And Evaluation Of Five Short-Term Load Forecasting
Techniques," IEEE Transaction on Power System, Vol.4, pp.1484-1491,1989.

[14] T. Haida and S. Muto, “Regression Based Peak Load Forecasting Using a Transformation
Technique,” IEEE Transaction on Power System, Vol.9, pp.1788–1794, 1994.

[15] D. Srinivasan and M. A. Lee, “Survey of Hybrid Fuzzy Neural Approaches to Electric Load
Forecasting," Proceedings of the IEEE International Conference on Systems, Man and Cybernetics,
Vancouver, BC, Part 5, pp.4004-4008, 1995.

[16] Alfares, H. & Nazeeruddin, M.. “Electric Load Forecasting: Literature Survey and Classification of
Methods“. International Journal of Systems Science. (33)1, pp. 23-34, 2002
[17] Almeshaiei, E., & Soltan, H. “ A methodology for Electric Power Load
Forecasting. “ Alexandria Engineering Journal, 50, pp.137–144.2011