Data Intelligence Just Accepted MS. - Ricerca sull'intelligenza artificiale specializzata al MIT

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

Total electricity consumption forecasting based on temperature

composite index and mixed-frequency models

LI Xuerong1,2, SHANG Wei1,2, *1, ZHANG Xun1, SHAN Baoguo3, WANG Xiang3

(1. Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing 100190, China;
2. MOE Social Science Laboratory of Digital Economic Forecasts and Policy Simulation at UCAS;

3. State Grid Energy Research Institute CO., LTD, Beijing 102209, China)

Abstract The total electricity consumption (TEC) can accurately reflect the operation of
the national economy, and the forecasting of the TEC can help predict the economic
development trend, as well as provide insights for the formulation of macro policies.
Nowadays, high-frequency and massive multi-source data provide a new way to predict
the TEC. in questo documento, a “seasonal-cumulative temperature index” is constructed based
on high-frequency temperature data, and a mixed-frequency prediction model based on
multi-source big data (Mixed Data Sampling with Monthly Temperature and Daily
Temperature index, MIDAS-MT-DT) is proposed. Experimental results show that the
MIDAS-MT-DT model achieves higher prediction accuracy, and the “seasonal-
cumulative temperature index” can improve prediction accuracy.

Key words Total electricity consumption; seasonal effect; temperature big data; high-
frequency big data; mixed-frequency prediction model

1 introduzione

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
D
N

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

Since electric power is closely related to industrial production, business activities

and residents’ living, electricity data could generally reflect the operation condition of the

national economy. Electricity statistics are of great value to be explored, which can help

the government to formulate macro-control policies and promote governance capacity

to look forward the economic or social development.

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

1* Corresponding author: Wei Shang, E-mail address: shangwei@amss.ac.cn (W. Shang).

This work was supported by the science and technology project of State Grid Corporation of China (Project Code:
1400-202157207UN-0-0-00); the National Natural Science Foundation of China [grant numbers 72273137].

© 2023 Chinese Academy of Sciences. Published under a Creative Commons Attribution 4.0
Internazionale (CC BY 4.0) licenza.

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

Among the statistical indicators of electricity, total electricity consumption (TEC)

is one of the most comprehensive and basic indicators to reflect the electricity

consumption situation of a country or region. TEC is generally defined as the total

electricity consumption of the primary, secondary and tertiary industries of the country

or region, including

industrial electricity, agricultural electricity, commercial

electricity, residential electricity, public facilities electricity, eccetera. The important value

of TEC lies in that it could reflect the operation condition of the national economy.

Accurate prediction of TEC can help track the trend of economic development and

provide insights for macro policymaking.

Tuttavia, the prediction of TEC is a difficult task, and there are few studies in

related fields. TEC includes various sectors of electricity consumption with different

patterns. Therefore, it is difficult to distinguish the complex factors influencing each

other during forecasting, which adds uncertainty to the prediction results. With the

development of big data, high-frequency big datasets that can reflect the micro behavior

of electricity consumption provide a new idea for the prediction of TEC. At present,

the existing researches in related fields mostly focus on the prediction of electricity load

[1-3], but there are still few models that can effectively predict TEC by multi-source

big datasets.

In this paper, a mixed-frequency prediction method based on temperature

composite index and mixed-data sampling (MIDAS) modello, MIDAS-MT-DT model,

is proposed and applied to TEC prediction, which significantly improves the prediction

accuracy compared with the benchmark models. Based on analyzing the electricity

consumption behavior in different seasons, this paper constructs the “seasonal-

cumulative temperature index”, which can more accurately reflect the electricity

consumption behavior affected by temperature. In addition, a high-frequency daily

TEC indicator is also introduced into the model to capture other factors except for

temperature. In order to simultaneously utilize the above two kinds of high-frequency

big data, we propose a mixed-frequency prediction model (MIDAS-MT-DT) for TEC

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
D
N

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

based on the “season-cumulative temperature index”, and select TEC of Fujian

province, China as the sample for empirical research. Through a series of comparative

experiments with benchmark models, it is verified that the MIDAS-MT-DT model has

higher prediction accuracy, and the “seasonal-cumulative temperature index” has the

ability to improve the prediction accuracy. The robustness and superiority of the

proposed framework are further verified by comparing it with more benchmark models

and multiple time windows.

The main contributions of this paper are in the following aspects: first, we put

forward a new perspective of constructing a temperature composite index to predict

electricity data. Most previous studies have selected a few specific months (summer or

winter) and used temperature data to predict local sample intervals [4-6]. IL

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
D
N

temperature index constructed in this paper involves all seasons in a unified analytical

framework, which is more compatible and helpful to reduce the application cost of the

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

actual system. Inoltre, we extend the traditional MIDAS model by incorporating

multi-frequency exogenous variables, thus improving the prediction ability of the

original model.

The remaining contents of this paper are arranged as follows: Sezione 2 summarizes

the existing literature on electricity consumption prediction and the mixed-frequency

modello; Sezione 3 presents the construction of a “seasonal-cumulative temperature

index”; Sezione 4 introduces the mixed frequency TEC forecasting model based on

temperature index. Sezione 5 compares the forecasting results of the models. Sezione 6

is a summary and outlook.

2 Literature review

In terms of electricity consumption analysis and prediction, there are many

forecasting methods proposed by scholars worldwide, which can be roughly divided

into classical forecasting methods, traditional forecasting methods and modern

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

intelligent forecasting methods. Among them, the classical prediction method includes

the elastic coefficient method, the calculation of the capacity of the expansion of the

industry, eccetera. The data frequency of traditional prediction methods is mainly annual and

monthly. The commonly used models include time series models [1], regression models

[2] gray prediction models [3-4], eccetera. With the development of data processing ability,

modern intelligent models have been widely used in electricity consumption

forecasting with monthly and daily basis data. Scholars have employed various models

including the neural network prediction method [5], support vector machine [6], chaos

theory prediction method [7], also include other combination forecast methods, eccetera.

In recent years, big data technology has gradually been applied in the research of

electricity data prediction [8-10]. The model of [11] found that for every 1-degree

increase in temperature, peak electricity consumption would increase by 0.45% A 4.6%.

Also using deep learning models, Bedi and Toshniwal (2020) propose a deep learning

based hybrid approach that firstly implements Variational Mode Decomposition (VMD)

and Autoencoder models to extract meaningful sub-signals/features from the data [12].

Ayub et al. (2020) applied the GRU-CNN model to predict the daily electricity

consumption of the ISO-NE data set, which improved the prediction accuracy by 7%

compared with the SOTA benchmark model [13]. Cui et al. (2023) propose a deep

learning framework with a COVID-19 adjustment for electricity demand forecasting

[14]. In the study of [15], the adaptive WT (AWT)-long short-term memory (LSTM) È

integrated into a hybrid approach for predicting electricity consumption.

Mixed-frequency models have initially been applied to the field of meteorology,

and the basic principle is to explore the information contained in the high-frequency

data and predict the future before the official release of relevant statistical data. IL

MIDAS model is a widely used mixed-frequency model, proposed by Ghysels et al.

(2004) [16]. Subsequently, many scholars have proposed extended forms of the MIDAS

modello, such as the MS-MIDAS model [17] and the co-integration MIDAS model [18].

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
D
N

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

More recently, the MF-VAR model has been applied to estimating the combined

endogenous variables [19].

Due to the demand for in-time forecasting in many industries, mixed-frequency

models have been applied to wind power forecasting, rail transit passenger flow

forecasting, macroeconomic forecasting, financial market forecasting and many other

fields. For example, some scholars applied multi-task learning and ensemble

decomposition methods to forecast wind power [20-22]. The in-time forecasting of

traffic ridership by Yao Enjian et al. (2018) and Bao Lei (2017) have greatly improved

the emergency response ability of the traffic system in emergencies [23, 24]. Currently,

mixed-frequency models are also widely used in macroeconomic and financial markets

[25-27]. Per esempio, Zhang Wei et al. (2020) [28] and Ghysel and Sinko (2011) [29]

respectively forecast Gross Domestic Product (GDP) and financial market volatility.

In summary, the existing literature applies various intelligent algorithms and

forecasts electricity data using historical datasets. Many studies have applied

meteorological big data or remote sensing big data and other natural environment data.

Most of the existing models use temperature data in specific months to forecast local

sample intervals but have not considered the multi-source high-frequency big data and

other predictive information, therefore the forecasting accuracy is expected to be further

improved.

3 Data and variables

This section presents the data collection and preprocessing, as well as the

construction of the “seasonal-cumulative temperature index”. Figura 1 presents the

methodology framework of the MIDAS-MT-DT model for TEC forecasting. IL

framework includes the following steps: 1) Collect the daily temperature data, daily

TEC data, monthly temperature data and monthly TEC data in the historical data; 2)

The daily temperature index is obtained by cumulative transformation and seasonal

transformation. The monthly temperature index is obtained by seasonal transformation.

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
D
N

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

3) The monthly temperature index and the lagged variable of the monthly TEC are

taken as the low-frequency forecasting variables, and the daily temperature index and

daily TEC are taken as the high-frequency forecasting variables. The low-frequency

and high-frequency variables are used to predict the monthly TEC. The details of the

above framework are presented in section 3 E 4.

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
D
N

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Figura 1. The framework of the MIDAS-MT-DT model

3.1 Data collection and preprocessing

In our empirical study, the monthly TEC of Fujian Province is selected as the

predicted variable. The sample period is from January 2017 to November 2020, and the

data source is the Wind database. The high-frequency big data used in the prediction

model includes 1) the daily TEC of Fujian Province, which is provided by State Grid

Energy Research Institute Co., LTD.; 2) The temperature data, which includes Xiamen,

Putian, Fuzhou, Nanping, Quanzhou, Ningde, Longyan, Sanming and Zhangzhou of

Fujian Province, and the data source is Wind database. The average of daily maximum

temperature on 9 cities of Fujian is taken as the original daily temperature data of the

temperature index (𝑇𝑖,𝑚 in eq. (1)); the average of the monthly average temperature on

9 cities of Fujian is taken as the original monthly temperature data of the temperature

index (𝑇𝑚 in eq. (3)). In the out-of-sample forecasting, we first use ARMA model to

predict temperature index on testing periods, and then the predicted values of

temperature index are inputted into our forecasting models. The sample period of the

Data collection and preprocessingThe construction of “Seasonal-cumulative temperature index”Mixed-frequency forecasting modelMonthly TECDaily temperature dataSeasonal transformationCumulative transformationMonthly temperature dataLagged daily TECMonthly temperature indexDailytemperatureindexHigh-frequency predicting variablesLow-frequency predicting variablesPredicted variableLagged monthly TECDaily TECHigh-frequency big data

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

above high-frequency data is from January 1, 2017, to November 30, 2020.

3.2 Seasonal-cumulative temperature index

Combined with seasonal changes, it can be analyzed that the electricity

consumption behavior has the following two characteristics: (1) Seasonal effect: Quando

the temperature is higher than the comfortable temperature, industrial production,

commercial and residential sectors need to use air conditioning to cool down. Al

same time, because Fujian province is in the southern region of China when the

temperature is lower than the comfortable temperature in winter, it also needs to use air

conditioning for heating. In summary, summer temperature should be positively

correlated with electricity consumption, while winter temperature should be negatively

correlated with electricity consumption. (2) Cumulative effect: the behavior of

electricity consumption has a certain inertia. The behavior of using air conditioning in

the first few days tends to continue for a short time, so the temperature of the first day

has a certain impact on the next few days.

Based on the two characteristics, the daily temperature data is transformed through

cumulative transformation and seasonal transformation. The formula of cumulative

transformation is:

4
(1 − ∑ 𝑒−𝑗
𝑗=1

𝑖

) ∗ 𝑇𝑖,𝑚 + ∑ 𝑒−𝑗

∗ 𝑇𝑖−𝑗,𝑚, 𝑖 ≥ 5

𝑗=1

4−𝑖

) ∗ 𝑇𝑖,𝑚 + ∑ 𝑒−𝑗

∗ 𝑇𝑖−𝑗+1,𝑚 + ∑ 𝑒−(4−𝑗)

∗ 𝑇𝑁−𝑖+1,𝑚−1, 𝑖 = 1,2,3,4

𝑗=1

𝑗=𝑖

𝐶𝑇 𝑖,𝑚 =

4
(1 − ∑ 𝑒−𝑗
𝑗=1

{

（1）

where 𝐶𝑇𝑖,𝑚 is the daily temperature index of the i day of the m month transformed by
cumulative effect; 𝑇𝑖,𝑚 is the original daily temperature data of the i day of the m

month; N indicates the total number of days in the m-1 month. j represents the number

of days before i. In our study, we assume the cumulative effect of electricity

consumption behavior lasts 5 days, thus 𝑗 ∈ {1,2,3,4}; 𝑒−𝑗 represents the influence

coefficient of the temperature of the previous j day, which decreases by the trend of

natural logarithm over time.

After that, the daily temperature data is transformed by seasonal effect

transformation, wherein the formula is:

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
D
N

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

𝑆𝐶_𝑇𝑖,𝑚 = {

𝐶𝑇𝑖,𝑚, 𝑚 ∈ {5,6,7,8,9}, 𝑖 = 1,2, … 𝑁
𝐿 − 𝐶𝑇𝑖,𝑚, 𝑚 ∈ {10,11,12,1,2,3,4}, 𝑖 = 1,2, … 𝑁

（2）

Dove, 𝑆𝐶_𝑇𝑖,𝑚 is the daily temperature index of the i day of the m month after

cumulative effect transformation and seasonal effect transformation, and N is the total

number of days of the m month. L is the displacement length to ensure that the

temperature index after seasonal transformation remains continuous. in questo documento, we

estimate the value of L by computing the average of dist( −𝐶𝑇𝑖,4, 𝐶𝑇𝑖,5 )+
dist(𝐶𝑇𝑖,9, −𝐶𝑇𝑖,10) of each year in the sample period, where dist(𝑎, 𝑏) represents the
distance between a and b, questo è, dist(𝑎, 𝑏)=|a-b|.

Inoltre, the monthly temperature data is transformed by seasonal effect to

obtain the monthly temperature index, in which the formula of seasonal effect

transformation is:

𝑆_𝑇𝑚 = {

𝑇𝑚, 𝑚 ∈ {5,6,7,8,9}
𝐿 − 𝑇𝑚, 𝑚 ∈ {10,11,12,1,2,3,4}

（3）

where 𝑆_𝑇𝑚 is the monthly temperature index of the m month after seasonal effect

transformation; 𝑇𝑚 is the original monthly temperature data of the m month; L is the

same as defined in eq. (4).

4 Methodology

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
D
N

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

In this section, the MIDAS-MT-DT mixed-frequency TEC forecasting model is

presented, as well as the single-frequency TEC forecasting models used for benchmark

models. After that, we present our experimental design of comparative experiments.

4.1 Mixed-frequency forecasting model based on multi-source big data

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

In order to comprehensively utilize high-frequency temperature data and high-

frequency electricity consumption data, the MIDAS model of Ghysels et al. (2004) È

extended in this paper, and a mixed-frequency prediction model of TEC (MIDAS-MT-

DT) based on “season-cumulative temperature index” is proposed. The formula of the

model is as follows:

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

𝑝𝑌−1

𝑝𝑋−1

𝑌𝑀,𝑡+1 = 𝜇 + ∑ 𝜇𝑗+1𝑌𝑡−𝑗

+ ∑ 𝛽𝑗+1𝑆_𝑇𝑡−𝑗

𝑝𝑋−1

𝑁−1

𝑗=0

𝑝𝑋−1

𝑁−1

+𝛽 ∑ ∑ 𝑤𝑁−𝑖+𝑗∗𝑁(𝜽)𝑌𝐷, 𝑁−𝑖,𝑡−𝑗

+ 𝛿 ∑ ∑ 𝑤𝑁−𝑖+𝑗∗𝑁(𝜽)𝑆𝐶_𝑇𝑁−𝑖,𝑡−𝑗

+ 𝑢𝑡+1

𝑗=0

𝑖=0

𝑗=0

𝑖=0

（4）

Dove, 𝑌𝑀,𝑡+1 is the monthly TEC of the t month, 𝑆_𝑇𝑡 is the monthly temperature

index of the t month after seasonal transformation, 𝑌𝐷, 𝑁,𝑡 is the daily TEC of the N day

of the t month, 𝑆𝐶_𝑇𝑁,𝑡is the daily temperature index of the N day of the t month after

cumulative transformation and seasonal transformation. 𝑤𝑖(𝜽) is a high-frequency

variable lag weighting polynomials of MIDAS model, 𝜽 is the estimated parameter

polynomial, and ∑

𝑝𝑋−1
𝑗=0

∑

𝑁−1
𝑖=0

𝑤𝑁−𝑖+𝑗∗𝑁(𝜽)

= 1 ; 𝜇 , 𝜇𝑗+1 , 𝛽 , 𝛽𝑗+1 , 𝛿 are the

parameters to be estimated by the model, 𝑝𝑋 and 𝑝𝑌 are the optimal lag period of the

model selected by AIC criterion, i and j are the integers in the summative operator, E

𝑢𝑡+1 is the random error of the model. The parameters are estimated by Non-linear

Least Squares (NLS).

In the comparison experiment, we employ Almon and Beta lag weight function of

MIDAS model, which are defined as

𝑤𝑖

𝐴𝑙𝑚𝑜𝑛(𝜃1, 𝜃2) =

𝑒𝜃1𝑖+𝜃2𝑖 2
∑ 𝑒𝜃1𝑖+𝜃2𝑖 2
𝑁
𝑖=1

𝑤𝑖

𝐵𝑒𝑡𝑎 (𝜃1, 𝜃2, 𝜃3) =

𝜃1−1(1 − 𝑥𝑖)𝜃2−1
𝑥𝑖
𝑁
∑ 𝑥𝑖
𝑖=1

𝜃1−1(1 − 𝑥𝑖)𝜃2−1

+ 𝜃3

(5)

(6)

.
where 𝑥𝑖 = (𝑖 − 1) (𝑁 − 1)

⁄

4.2 Single-frequency forecasting models

In order to verify the predictive power of the MIDAS-MT-DT model and the

“seasonal-cumulative temperature index”, several comparative experiments of single-

frequency forecasting models are conducted. Primo, daily and monthly Autoregressive

Moving Average (ARMA) models are used to compare the prediction accuracy with

and without the temperature index. Specifically, the monthly ARMA model is:

𝑝𝑌−1

𝑝𝑋−1

𝑝𝑍−1

𝑌𝑀,𝑡+1 = 𝜇 + ∑ 𝜇𝑗+1𝑌𝑀,𝑡−𝑗

+ ∑ 𝛽𝑗+1𝑆_𝑇𝑡−𝑗

+ ∑ 𝛿𝑗+1𝑢𝑡+1

（7）

𝑗=0

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
D
N

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

where 𝑌𝑀,𝑡 is the monthly TEC in month t, 𝑆_𝑇𝑡is the monthly temperature index of

the t month after seasonal transformation, and 𝑢𝑡+1 is an independent and identically

distributed random variable, representing the model error.

The daily ARMA model is:

𝑝𝑌−1

𝑝𝑋−1

𝑝𝑍−1

𝑌𝐷,𝑡+1 = 𝜇 + ∑ 𝜇𝑗+1𝑌𝐷,𝑡−𝑗

+ ∑ 𝛽𝑗+1𝑆𝐶_𝑇𝑡−𝑗

+ ∑ 𝛿𝑗+1𝑢𝑡+1

（8）

𝑗=0

Dove, 𝑌𝐷,𝑡 is the daily TEC on the day t, and 𝑆𝐶_𝑇𝑡 is the daily high-frequency

temperature index on the day t after seasonal and cumulative transformation.

4.3 Experimental design

The prediction accuracy of MIDAS models with daily or monthly temperature

index and without temperature index are compared respectively. Inoltre, mixed-

frequency models are compared with several single-frequency models. In addition to

ARMA models, some intelligent models such as Support Vector Regression (SVR) E

Random Forest (RF) model are selected as the benchmark models. To sum up, IL

model specifications and parameters of the comparison experiments are shown in Table

Tavolo 1. Model specifications and parameters

Model label

Model specifications

Model Parameters

ARMA-M

Single-frequency monthly ARMA model, con

In eq. (7), 𝛽𝑗 = 0

lagged variables and without temperature index

ARMA-D

Single-frequency daily ARMA model, con

In eq. (8), 𝛽𝑗 = 0

lagged variables and without temperature index

ARMA-MT

Monthly ARMA model with monthly

In eq. (7), 𝛽𝑗 ≠ 0

temperature index and lagged variables

ARMA-DT

Daily ARMA model with daily temperature

In eq. (8), 𝛽𝑗 ≠ 0

index and lagged variables

SVR-M

Single-frequency monthly SVR model, con

lagged variables and without temperature index

SVR-D

Single-frequency daily SVR model, con

lagged variables and without temperature index

Default values of Python Scikit-

SVR-MT

Monthly SVR model with monthly temperature

learn library

index and lagged variables

SVR-DT

Daily SVR model with daily temperature index

and lagged variables

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
D
N

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

RF-M

Single-frequency monthly RF model, con

lagged variables and without temperature index

RF-D

Single-frequency daily RF model, with lagged

variables and without temperature index

RF-MT

Monthly RF model with monthly temperature

index and lagged variables

RF-DT

Daily RF model with daily temperature index

and lagged variables

MIDAS

Mixed-frequency benchmark model, without

In eq. (4), 𝛽𝑗 = 0，𝛿 = 0

temperature index

MIDAS-MT

MIDAS model with monthly temperature index

In eq. (4), 𝛽𝑗 ≠ 0，𝛿 = 0

and its lagged variables

MIDAS-DT

MIDAS model with daily temperature index

MIDAS-MT-DT

MIDAS model with monthly temperature index

In eq. (4), 𝛽𝑗 = 0，𝛿 ≠ 0
In eq. (4), 𝛽𝑗 ≠ 0，𝛿 ≠ 0

and its lagged variables, as well as the daily

temperature index

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

5 Empirical results

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

e
D
tu
D
N

In this section, the forecasting performances of the MIDAS-MT-DT model and the

“seasonal-cumulative temperature index” are illustrated through a series of comparative

experimental results. Primo, the description of the “seasonal-cumulative temperature

index” are presented, and then the prediction results of the single-frequency models and

the mixing-frequency models are compared respectively.

5.1 Data description and correlation analysis

According to the construction method in Section 3.2, the description of the

“season-cumulative temperature index” is shown in Figure 2. According to the results

in the figure, the temperature index constructed in this paper maintains a general trend

of positive correlation with Fujian TEC, which indicates that it may improve the

forecasting performance of temperature data.

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

Raw temperature data

Seasonal-cumulative temperature index

Fujian TEC

Figura 2. The description of “season-cumulative temperature index”

Tavolo 2 shows the descriptive statistics and correlation analysis results of the raw

temperature data, “seasonal-cumulative temperature index” and daily Fujian TEC. IL

results show that the correlation coefficient between the “seasonal-cumulative

temperature index” and Fujian TEC is 0.6540, while the correlation coefficient between

the raw temperature data and Fujian TEC is only -0.3060. This indicates that the

temperature index construction method in this paper can effectively improve the

correlation with the predicted variables.

Tavolo 2. Descriptive statistics and correlation analysis

Raw temperature data

Seasonal-cumulative

Fujian TEC

temperature index

Mean

Maximum

Minimum

Std. Dev.

Skewness

Kurtosis

Pearson correlation

coefficient with Fujian TEC

26.1478

9.4058

37.2591

6.5261

-1.0306

-0.2506

-0.3060

35.5638

50.5942

22.3403

5.1441

0.2943

2.6980

0.6540

6.2101

3.4423

8.9278

0.9815

0.2392

-0.0345

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
D
N

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

5.2 Single-frequency forecasting models

To verify the predictive ability of the “seasonal-cumulative temperature index”,

daily and monthly models are used to compare the prediction accuracy with and without

the temperature index, rispettivamente. The prediction results are shown in Table 3. Nel

table, column 1 represents the testing period, and the corresponding training period is

from the beginning of the sample to the previous month of the testing period. IL

prediction accuracies are calculated by the following formulas:

𝐴𝑐𝑐𝑢𝑟𝑎𝑐𝑦 (𝐴𝐶𝐶) = 1 −

1
𝑁

𝑁

∑ |
𝑡=1

𝑥̂𝑡 − 𝑥𝑡
𝑥𝑡

𝑅𝑜𝑜𝑡 𝑀𝑒𝑎𝑛 𝑆𝑞𝑢𝑎𝑟𝑒𝑑 𝐸𝑟𝑟𝑜𝑟 (𝑅𝑀𝑆𝐸) = √

1
𝑁

𝑁

∑(
𝑡=1

𝑥̂𝑡 − 𝑥𝑡
𝑥𝑡

（7）

（8）

where the 𝑥̂𝑡 and 𝑥𝑡 represent the predicted value and the real value of the forecast

modello, rispettivamente.

Tavolo 3. Prediction results of single-frequency models

Panel A: Monthly frequency models

Testing period

ARMA-M

ARMA-MT

SVR-M

SVR-MT

RF-M

RF-MT

2020.1-

2020.3

2020.4-

2020.6

2020.7-

2020.9

ACC

63.42%

83.20%

76.84%

92.67%

77.60%

92.85%

RMSE

17.7056

16.0453

23.9215

7.2015

24.1843

7.9507

ACC

62.89%

70.12%

87.11%

93.11%

86.54%

91.71%

RMSE

54.6691

53.8693

27.9021

14.8238

20.8601

15.2980

ACC

62.43%

82.31%

87.48%

89.13%

81.16%

86.97%

RMSE

58.1760

33.7527

23.9188

21.8095

31.4153

25.0699

2020.10-

ACC

69.75%

75.52%

74.02%

89.64%

73.95%

85.67%

2020.11

RMSE

48.1106

33.5277

39.9282

16.1358

34.1574

21.0209

Panel B: Daily frequency models

Testing period

ARMA-D

ARMA-DT

SVR-D

SVR-DT

RF-D

RF-DT

2020.8.1-

ACC

61.10%

97.72%

82.92%

95.49%

85.16%

94.98%

2020.8.31

RMSE

3.1863

0.2353

1.6579

0.4335

1.4492

0.4886

2020.9.1-

ACC

67.51%

89.07%

84.03%

95.76%

87.21%

88.73%

2020.9.30

RMSE

2.5071

0.9195

1.4930

0.3945

1.2188

0.9456

2020.10.1-

ACC

76.72%

96.39%

74.48%

95.24%

82.61%

93.19%

2020.10.31

RMSE

1.5694

0.3934

1.9787

0.3897

1.3973

0.5174

2020.11.1-

ACC

74.17%

98.23%

84.62%

95.22%

79.36%

94.30%

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
D
N

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

2020.11.30

RMSE

1.7447

0.1587

1.3304

0.3790

1.6610

0.4516

Note: The bold numbers in the table indicate models with improved predictive accuracy compared

to the benchmark model.

According to the results in the table, in all testing periods, regardless on the daily

or the monthly basis, the prediction accuracies of forecasting models added with the

temperature index are significantly improved compared with the benchmark models.

Intelligent models such as SVR and RF perform much better than ARMA models in the

cases of the monthly models. On the daily basis, intelligent models and ARMA models

are competitive. Among them, the highest prediction accuracy has reached 98.23%. IL

results in Table 3 show that the temperature index constructed in this paper can

significantly improve the forecasting ability of the benchmark models by accurately

reflecting the influence of electricity consumption behavior on TEC.

5.3 Mixed-frequency forecasting models

In order to verify the prediction ability of the MIDAS-MT-DT model, IL

prediction accuracy of the MIDAS model with daily, and monthly temperature index

and without temperature index are compared respectively. The prediction results are

shown in Table 4. The first column in the table represents the testing period of the model,

and the corresponding training period is from the beginning of the sample period to the

previous month of the testing period. The prediction accuracies of the corresponding

models are calculated by eq. (7-8).

Tavolo 4. Prediction results of mixed-frequency models

Panel A: Total Electricity Consumption of Fujian

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
D
N

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Almon-MIDAS

Beta-MIDAS

Testing period

MIDAS

MIDAS-MT MIDAS-DT MIDAS-MT-DT

MIDAS MIDAS-MT MIDAS-DT MIDAS-MT-DT

2020.1-
2020.3

2020.4-
2020.6

2020.7-
2020.9

2020.10-
2020.11

ACC

RMSE

ACC

RMSE

ACC

RMSE

ACC

RMSE

83.21%

86.15%

16.6583

21.0858

96.54%

93.35%

4.8878

7.2637

94.45%

94.91%

9.0774

7.5643

94.23%

95.62%

90.86%

10.4433

94.00%

9.0697

94.30%

7.9832

89.58%

5.3534

4.2126

12.2389

86.15%

12.4566

98.39%

2.3340

96.29%

6.1133

95.18%

3.5138

80.68%

94.30%

81.87%

18.8926

5.0649

18.6801

84.34%

91.62%

90.94%

23.8644

14.2741

14.4594

93.64%

91.13%

87.83%

11.5742

17.2138

20.4678

90.66%

93.09%

93.38%

13.0001

12.4452

8.8826

96.93%

3.4150

94.72%

7.3290

96.73%

5.1104

98.02%

3.7987

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

Panel B: Residential Electricity Consumption of Fujian

Almon-MIDAS

Beta -MIDAS

Testing period

MIDAS

MIDAS-MT MIDAS-DT MIDAS-MT-DT

MIDAS MIDAS-MT MIDAS-DT MIDAS-MT-DT

2020.1-
2020.3

2020.4-
2020.6

2020.7-
2020.9

2020.10-
2020.11

ACC

RMSE

ACC

RMSE

ACC

RMSE

ACC

RMSE

91.51%

94.84%

3.4390

2.1221

90.57%

3.8354

88.79%

87.20%

90.74%

5.4743

5.6525

88.39%

94.68%

7.8735

4.4300

4.7849

86.43%

9.9840

87.59%

92.30%

95.49%

5.0707

3.5147

1.7105

92.69%

4.1736

94.96%

2.2653

94.59%

3.1155

93.74%

2.5531

80.89%

85.07%

84.96%

16.4449

14.7521

15.5679

88.86%

92.04%

83.15%

20.1651

11.3245

26.7171

82.40%

91.79%

92.87%

26.8130

11.6818

10.3671

86.68%

85.01%

90.50%

21.6887

26.0945

13.6509

Panel C: The Tertiary Industry Electricity Consumption of Fujian
Almon-MIDAS

Beta -MIDAS

92.33%

9.0528

95.59%

7.5793

94.95%

11.6454

92.23%

13.8566

D
o
w
N
o
UN
D
e
D

Testing period

MIDAS

MIDAS-MT MIDAS-DT MIDAS-MT-DT

MIDAS MIDAS-MT MIDAS-DT MIDAS-MT-DT

2020.1-
2020.3

2020.4-
2020.6

2020.7-
2020.9

2020.10-
2020.11

ACC

RMSE

ACC

RMSE

ACC

RMSE

ACC

RMSE

82.26%

91.45%

4.8074

2.7748

91.48%

93.18%

3.0102

1.8038

81.06%

4.4253

85.19%

4.0636

90.77%

89.33%

93.59%

5.3767

5.5239

95.12%

90.61%

2.0203

3.9438

3.2686

90.73%

3.2524

94.32%

1.6667

94.57%

1.4146

97.24%

1.3215

97.25%

0.9801

84.00%

83.56%

78.24%

4.1694

4.7109

4.7102

84.77%

81.86%

89.79%

5.1318

5.9643

3.3482

77.35%

90.60%

82.25%

11.3417

4.8776

7.9887

83.82%

90.98%

79.14%

5.93041

3.0239

7.1812

93.40%

2.5181

91.51%

3.0738

86.34%

6.7196

91.15%

3.2041

Note: The bold numbers in the table indicate models with improved predictive accuracy compared to the benchmark MIDAS model.

According to the results in the table, in all testing periods, regardless of whether

the daily temperature index or monthly temperature index is added, the prediction

accuracies of the models are significantly improved compared with the benchmark

models. Among them, the highest prediction accuracy has reached 98.39%. Inoltre,

the MIDAS-MT model and MIDAS-MT-DT model have obtained higher prediction

accuracies than the benchmark models in most of the four testing periods. Tuttavia,

only one of the MIDAS-DT models achieves higher prediction accuracy. This indicates

that the monthly temperature index has a better predictive ability than the daily

temperature index. By comparing different lag weighting polynomials of MIDAS

models, we also find that Almon-MIDAS models perform better that Beta-MIDAS on

most of the data samples.

To verifying the robustness of our model on more datasets, we further apply our

models to forecast Residential Electricity Consumption (REC) of Fujian and The

Tertiary Industry Electricity Consumption of Fujian. The results in Panel B and C of

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
D
N

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

Tavolo 4 show that MIDAS-MT-DT model achieves better performances on all testing

periods of multiple datasets. We have also observed similar results as those in Panel A

Quello, the monthly temperature index has a stronger ability to improve the prediction

accuracy in the mixing-frequency model. Overall, our results demonstrate that the

MIDAS-MT-DT model proposed in this paper significantly improves the prediction

accuracy of the TEC by incorporating high-frequency temperature data and high-

frequency TEC data, and this advantage is not easily affected by the randomness of the

data set with good robustness.

6 Conclusions and future research

The total electricity consumption (TEC) reflects the operation of the national

economy. Accurate prediction of the TEC is of great significance for the country to look

forward the economic development and formulate macro-control policies. The high-

frequency and massive multi-source data provides a new idea for the prediction of TEC.

Based on the analysis of electricity consumption behavior in different seasons, Questo

study constructs a “seasonal-cumulative temperature index” considering the inertia of

seasons and electricity consumption behavior, which can reflect the electricity

consumption behavior affected by temperature. Inoltre, high-frequency daily data

of the TEC is also incorporated to supplement the electricity consumption behavior

affected by other factors. Based on the above two high-frequency datasets, this study

proposes a mixed-frequency prediction model (MIDAS-MT-DT) for the TEC based on

the “seasonal-cumulative temperature index” and mixed-frequency models.

According to the empirical results, the temperature index constructed in this paper

is able to significantly improve the forecasting ability of the benchmark model by

reflecting the electricity consumption behavior affected by temperature and other

factors. By incorporating high-frequency temperature data and daily TEC data, IL

MIDAS-MT-DT model proposed in this paper captures the intricate factors of

electricity consumption behavior, thus significantly improves the prediction accuracy

of the TEC, and the highest accuracy has reached 98.39%. Through comparative

esperimenti, we find that the monthly temperature index has a stronger ability to

improve the prediction accuracy in the mixing-frequency model. The experiments of

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
D
N

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

multiple testing periods and predicted datasets further verify the robustness of the

MIDAS-MT-DT model.

In terms of the limitations of this paper, future research directions include: In

addition to the big data of temperature, other microscopic data that can reflect the

behavior of electricity consumption can be further collected, and more exogenous

variables can be introduced into the prediction model. Examples include remote sensing

data and internet data. In terms of the prediction model, the fusion technology of multi-

source electricity big data, machine learning and deep learning models can be explored

to further improve the prediction accuracy.

Author Contributions

Xuerong Li has collected and processed experimental data, presented experiment results and

drafted the original version of the manuscript; Wei Shang has guided the overall direction of

questa ricerca, and provided advice for comparison experiments. Xun Zhang provided advice for

motivation and conclusions presented in Section 1. Baoguo Shan and Xiang Wang have funded

the research, provided the proposal of the research, and contributed part of the experimental

dati. All the authors have made meaningful and valuable contributions in revising and

proofreading the resulting manuscript.

Riferimenti

[1] UN. Hussain, M. Rahman & J. UN. Memon. Forecasting electricity consumption in Pakistan: IL

way forward. Energy Policy 90 (2016), 73-80. doi: 10.1016/j.enpol.2015.11.028.

[2] Y. Feng & S. M. Ryan. Day-ahead hourly electricity load modeling by functional regression.

Applied Energy, 170 (2016), 455-465. doi: 10.1016/j.apenergy.2016.02.118.

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
D
N

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

[3] W. Xu, R. Gu, Y. Liu & Y. Dai. Forecasting energy consumption using a new GM–ARMA

model based on HP filter: The case of Guangdong province of China. Economic Modelling,

45 (2015), 127-135. doi: 10.1016/j.econmod.2014.11.011.

[4] l. Zeng, C. Liu & W. Z. Wu. A novel discrete GM (2, 1) model with a polynomial term for

forecasting electricity consumption. Electric Power Systems Research, 214 (2023), 108926.

doi: 10.1016/j.epsr.2022.108926.

[5] N. Kunwar, K. Yash & R. Kumar. Area-load based pricing in DSM through ANN and heuristic

scheduling. Smart Grid, 4 (2013), 1275-1281. doi: 10.1109/TSG.2013.2262059.

[6] Q. Cheng, Y. Yan, S. Liu, C. Yang, H. Chaoui & M. Alzaed. Particle filter-based electricity

load prediction for grid-connected microgrid day-ahead scheduling. Energies, 13 (2020),

6489. doi: 10.3390/en13246489.

[7] Z. J. Liu, H. M. Yang & M. Y. Lai. Electricity price forecasting model based on chaos theory.

In: Proceedings of 2005 International Power Engineering Conference, 2005, pag. 1-9. doi:

10.1109/IPEC.2005.206950.

[8] X. Guo, Q. Zhao, D. Zheng, Y. Ning & Y. Gao. A short-term load forecasting model of multi-

scale CNN-LSTM hybrid neural network considering the real-time electricity price. Energy

Reports, 6 (2020), 1046-1053. doi: 10.1016/j.egyr.2020.11.078.

[9] P. Jiang, Y. Nie, J. Wang & X. Huang. Multivariable short-term electricity price forecasting

using artificial intelligence and multi-input multi-output scheme. Energy Economics, 117

(2023), 106471. doi: 10.1016/j.eneco.2022.106471.

[10] Y. Jiang, T. Gao, Y. Dai, R. Si, J. Hao, J. Zhang & D. W. Gao. Very short-term residential

load forecasting based on deep-autoformer. Applied Energy, 2022, 328: 120120. doi:

10.1016/j.apenergy.2022.120120.

[11] M. Santamouris, C. Cartalis, UN. Synnefa, D. Kolokotsa. On the impact of urban heat island

and global warming on the power demand and electricity consumption of buildings—A

revisione. Energy & Buildings, 98 (2015), 119-124. doi: 10.1016/j.enbuild.2014.09.052.

[12] J. Bedi & D. Toshniwal. Energy load time-series forecast using decomposition and

autoencoder integrated memory network. Applied Soft Computing, 93 (2020), 106390. doi:

10.1016/j.asoc.2020.106390.

[13] N. Ayub, M. Irfan, M. Awais, U. Ali, T. Ali, M. Hamdi, UN. Alghamdi & F. Muhammad. Big

data analytics for short and medium term electricity load forecasting using AI techniques

ensembler. Energies, 13 (2020), 5193. doi: 10.3390/en13195193.

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
D
N

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

[14] Z. Cui, J. Wu, W. Lian & Y. G. Wang. A novel deep learning framework with a COVID-19

adjustment for electricity demand forecasting. Energy Reports, 9 (2023), 1887-1895. doi:

10.1016/j.egyr.2023.01.019.

[15] UN. Saranj & M. Zolfaghari. The electricity consumption forecast: Adopting a hybrid

approach by deep learning and ARIMAX-GARCH models. Energy Reports, 8 (2022), 7657-

7679. doi: 10.1016/j.egyr.2022.06.007.

[16] E. Ghysels, P. Santa-Clara P & R. Valkanov. The MIDAS touch: Mixed data sampling

regression models. UC

Los Angeles:

Finance.

(2004). Available

at:

http://escholarship.org/uc/item/9mf223rs.

[17] P. Guérin & M. Marcellino. Markov-switching MIDAS model. Journal of Business &

Economic Statistics, 31 (2013), 45-56. doi: 10.1080/07350015.2012.727721.

[18] J. IO. Mugnaio. Mixed-frequency cointegrating regressions with parsimonious distributed lag

structures. Journal of Financial Econometrics, 12 (2014), 584-614. doi: 10.1093/jjfinec/nbt010.

[19] R. Kikuchi, T. Misaka, S. Obayashi, H. Inokuchi, H. Oikawa & UN. Misumi. Nowcasting

algorithm for wind fields using ensemble forecasting and aircraft flight data. Meteorological

Applications, 25 (2018), 365-375. doi: 10.1002/met.1704.

[20] UN. Dupré, P. Drobinski P, J. Badosa, C. Briard & P. Tankov. The economic value of wind

energy nowcasting. Energies, 13 (2020), 5266. doi: 10.3390/en13205266.

[21] IO. Kutiev, P. Muhtarov, B. Andonov & R. Warnant. Hybrid model for nowcasting and

forecasting the K index. Journal of Atmospheric and Solar-Terrestrial Physics, 71 (2009), 589-

596. doi: 10.1016/j.jastp.2009.01.005.

[22] Y. Wei & M. C. Chen. Forecasting the short-term metro passenger flow with empirical mode

decomposition and neural networks. Transportation Research Part C Emerging Technologies,

21 (2012), 148-162. doi: 10.1016/j.trc.2011.06.009.

[23] M. Ni, Q. Lui & J. Gao. Forecasting the subway passenger flow under event occurrences with

social media. IEEE Transactions on Intelligent Transportation Systems, 18 (2017), 1623-1632.

doi: 10.1109/TITS.2016.2611644.

[24] V. Kuzin, M. Marcellino & C. Schumacher. MIDAS vs. mixed-frequency VAR: Nowcasting

GDP in the euro area. International Journal of Forecasting, 27 (2011), 529-542. doi:

10.1016/j.ijforecast.2010.02.006.

[25] E. Andreou, E. Ghysels, UN & Kourtellos. Should macroeconomic forecasters use daily

financial data and how?. Journal of Business & Economic Statistics, 31 (2013), 240-251. doi:

10.1080/07350015.2013.767199.

[26] F. Corsi. A simple approximate long-memory model of realized volatility. Journal of Financial

Econometrics, 7 (2009), 174-196. doi: 10.1093/jjfinec/nbp001.

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
D
N

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

[27] H. Bahcivan & C. C. Karahan. High frequency correlation dynamics and day-of-the-week

effect: A score-driven approach in an emerging market stock exchange. International Review

of Financial Analysis, 80 (2022), 102008. doi: 10.1016/j.irfa.2021.102008.

[28] UN. Richardson, T. Mulder & T. Vehbi. Nowcasting GDP using machine-learning algorithms:

A real-time assessment. International journal of forecasting, 37 (2021), 941-948. doi:

10.1016/j.ijforecast.2020.10.005.

[29] E. Ghysels & UN. Sinko. Volatility forecasting and microstructure noise. Journal of

Econometrics, 160 (2011), 257-271. doi: 10.1016/j.jeconom.2010.03.035.

Author Biography

Dr. Xuerong Li, assistant professor of Academy of Mathematics and

Systems Science, Chinese Academy of Sciences. She received the B.E.

degree from Renmin University of China and the Ph.D. degree from

University of Chinese Academy of Sciences in 2013 E 2019

rispettivamente. Her research interests are in the area of economic

forecasting and machine learning.

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
D
N

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3

Prof. Wei Shang, associate professor of Academy of Mathematics and

Systems Science, Chinese Academy of Sciences, chief engineer of

Center for Forecasting Science, Chinese Academy of Sciences. She

holds a PhD from Harbin Institute of Technology. Her research focuses

on macro-economic monitoring and early warning, Internet Data

Mining, e-business and information technology and development. She

published her researches in major journals such as Decision Support Systems, Electronic

Research and Applications, and Online Information Review. She is principle investigator of

three regular project of National Science Foundation of China, as well as an ICT4D project of

Information Development Research Center (Cananda).

Data Intelligence Just Accepted MS.
https://doi.org/10.1162/dint_a_002155

Prof. Xun Zhang, associate professor of Academy of Mathematics and

Systems Science, Chinese Academy of Sciences. She received the B.E.

degree from Renmin University of China and the Ph.D. degree from

Academy of Mathematics and Systems Science, Chinese Academy of

Sciences in 2004 E 2009 rispettivamente. Her research interests are in the

area of macroeconomic forecasting and energy economics.

Shan Baoguo is the vice president of State Grid Energy Research Institute

and a professor-level senior engineer. He received his bachelor’s degree and

master’s degree from North China Electric Power University in 1993 E

1997, rispettivamente. His research interests are energy power analysis and

forecasting, power demand side management.

Dr. Wang Xiang is a senior economist of State Grid Energy Research

Institute. He received his bachelor’s degree from Jilin University in 2011 E

his Ph.D.’s degree from Nankai University in 2014. His research interests

are macroeconomic and power market analysis and forecasting.

D
o
w
N
o
UN
D
e
D

F
R
o
M
H

T
T

:
/
/

D
io
R
e
C
T
.

io
T
.

e
D
tu
D
N

T
/

UN
R
T
io
C
e
–
P
D

F
/

D
o

io
/

1
0
1
1
6
2
D
N
_
UN
_
0
0
2
1
5
2
1
2
7
0
2
5
D
N
_
UN
_
0
0
2
1
5
P
D

B
sì
G
tu
e
S
T

o
N
0
7
S
e
P
e
M
B
e
R
2
0
2
3 Data Intelligence Just Accepted MS. Immagine

Scarica il pdf