Bridging Tradition and Modernity: The Application and Reflection

of Information Technology in Census Practices

Jiahui Chen

College of Art & Sciences, Department of Communication, University of Washington, Seattle, U.S.A.

Keywords: Population Forecasting, Demographic Models, Small-Area Projections, Machine Learning, Probabilistic

Methods.

Abstract: Countries and governments need to regularly understand population changes and predict population trends. This

is very important for planning housing, employment, health care and other public

services. Population

forecasting helps to reasonably and accurately allocate

resources to maximize utilization. Traditional

statistical models such as cohort distribution cannot explain population trends caused by global events such as

politics, economy, and immigration. LSTM networks provide a computational method for sustainable

observation to adapt to complex population patterns. Artificial intelligence and scientific statistics have made

predictions easier, but they still require a lot of data support. Population forecasting and census are more

complex issues. This paper studies the limitations and reliability of different measurement methods, and

considers other external factors such as climate, policy, and economic factors that affect population size, such as

real challenges. By studying and summarizing these factors, this study will emphasize the differences in

forecasting methods and consider how to improve the accuracy and adaptability of population forecasts and

censuses in the future.

1 INTRODUCTION

Being able

predict

population trends

is crucial

for governments, businesses, and planners. Without

accurate forecasts, cities might not build enough

schools, hospitals, or housing to

keep

up with

demand. Businesses also

rely on population

predictions to decide where to expand or what

products to offer. Governments use these

forecasts to

plan

long-term policies on infrastructure, healthcare,

and economic development.

For decades, most population predictions were

done using demographic models, like the cohort-

component method, which estimates future

populations based on birth rates, death rates, and

migration patterns(Aryal, 2020). These models work

well

for

large

areas

with stable trends but don't

always perform

as well

for smaller

regions or

places experiencing sudden changes, like economic

booms or

natural

disasters. That's why

researchers

have

been developing more advanced forecasting

methods that consider uncertainty, spatial variation,

and machine learning techniques(Yu et

al., 2023).

https://orcid.org/0009-0008-8576-5614

Population changes vary widely depending on

location. In some places, like South Korea, aging

populations are a major concern, and forecasting

helps governments

prepare for rising

healthcare

costs and pension demands (Kim &

Kim, 2020).

Meanwhile, in

fast-growing urban areas,

projections need to factor in high migration rates and

shifts in birth

patterns (Sang et

al., 2024). Because

of these differences, improving forecasting methods

is necessary to

ensure accurate predictions for

different regions and

circumstances.

There

exist a

number of

methodologies

employed in population projection with their

respective merits and demerits. Aryal utilized the

cohort-component method, one of the most popular

demographic approaches, to project future

populations by an analysis of historical data of birth

rates, death rates, and migration(Aryal, 2020). This

approach has shown itself to be trustworthy for

forecasting long- term trends, especially in settled

areas, but is frequently not able to manage unexpected

demographic disruptions or sudden shifts, particularly

those instigated by political or environmental crises.

Chen, J.

Bridging Tradition and Modernity: The Application and Reﬂection of Information Technology in Census Practices.

DOI: 10.5220/0013688000004670

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 2nd International Conference on Data Science and Engineering (ICDSE 2025), pages 285-291

ISBN: 978-989-758-765-8

285

help

alleviate the limitations

of single-point

projections, Vollset et

al. put

forward

a probabilistic

projection

framework

that

could produce

a range

possible population scenarios. The

study, which

included 195

countries and

territories, demonstrated

that

the

inclusion

of uncertainty

regarding

fertility

and migration substantially improves accuracy at

both the national and regional levels (Vollset, 2020).

Pointing to the difficulties with making

projections at more

localized

levels, Kim and Kim

and Sang et al. considered small-area and spatial

projection models

(Kim & Kim, 2020; Sang

et al.,

2024). These

models

synthesize localized trends and

geographic information to make improved

predictions

for individual municipalities

or counties,

especially where population trends are very different

from the national average.

In the last several years, scholars have

increasingly applied artificial intelligence to improve

population

projection. Grossman et

al. applied long

short-term

memory (LSTM)

neural

networks to

small-area forecasting and concluded

that

AI-based

models have the potential to outperform conventional

methods in

uncovering

intricate patterns

in big

data

(Grossman et al., 2023). Nevertheless, the success of

such

approaches hinges greatly

data

quality

and

availability, and the interpretability of

such

models

continues to challenge demographers and

policymakers.

This

paper will explore different

population

forecasting techniques and compare their

effectiveness. The main areas of focus include:

1. How traditional demographic models compare

to newer probabilistic and AI-driven

techniques.

2. The challenges in forecasting smaller

populations and improving prediction

accuracy at the local level.

3. The

potential for

to enhance population

forecasting and whether it can work alongside

traditional models.

By analyzing these

methods, this paper aims to

provide insight into the best approaches

for different

forecasting needs and how they can be refined. Since

reliable population predictions

are necessary

for

effective decision-making, improving forecasting

techniques will remain an important area of research.

2 RESEARCH METHODS AND

FORECASTING MODELS

Population forecasting has evolved significantly over

the

past decades, expanding from deterministic

demographic models to probabilistic approaches and

artificial intelligence (AI)-driven algorithms. This

section systematically reviews three main

methodological categories:

traditional cohort-based

projections, probabilistic and spatial forecasting

techniques, and

machine learning-based models.

Each has unique strengths and weaknesses depending

on the

forecasting

scale, data availability, and policy

applications.

2.1 Cohort-Based and Structured

Demographic Models

Traditional

demographic

forecasting often begins

with the cohort-component

model, which

segments

the

population by age and sex and

applies

projected

fertility, mortality, and migration rates

to simulate

future changes. Aryal outlines the

foundational

role

this method plays in national statistics offices, noting

its relative simplicity and strong performance in

contexts with stable demographic trends (Arya, 2020).

It is widely adopted due to its transparency and

compatibility with census data.

However, its limitations have become

increasingly apparent, particularly in cases where

rapid change

or local

heterogeneity undermines the

assumption of linearity. Ellner et al. expand on these

structural models by

comparing

integral projection

models (IPMs) and matrix population models

(MPMs), arguing that while

both

frameworks

offer

mathematical precision, they assume stable

environments and are thus ill-equipped to model

abrupt

demographic shifts caused by unexpected

events, such as

the COVID-19

pandemic

or sudden

policy shifts (Ellner et al., 2022).

Kim and Kim highlighted these limitations in their

sub-national study of South Korea’s aging population.

Their research

revealed that

while

national-level

projections

often

suggest moderate

trends, local

regions

experience much sharper demographic

transitions, requiring forecasting methods

that

account for age-specific dynamics and spatial

differentiation (Kim & Kim, 2020).

Overall, while

structured demographic models

remain a cornerstone in official forecasting, their

reliance on historic trend

stability, and

their limited

adaptability

to granular regional differences, limits

ICDSE 2025 - The International Conference on Data Science and Engineering

286

their utility in

increasingly volatile demographic

landscapes.

2.2 Probabilistic Forecasting and

Spatial Models

contrast to

deterministic

models, probabilistic

forecasting techniques explicitly account for

uncertainty in demographic variables. Rather than

offering a single-point estimate, these models

produce distributions of outcomes based on

simulations or Bayesian inference frameworks.

Vollset et al. introduced a comprehensive

probabilistic approach for

195 countries in the Global

Burden of

Disease study, modeling population

scenarios

from

2017 to

2100 (Vollset

et al., 2020).

Their

framework

incorporates

stochastic

variation

fertility, mortality, and migration

assumptions,

yielding more

informative

forecasts, especially

for

long-term planning.

Yu et al. extend probabilistic modeling to the

county level in the U. S., emphasizing that small-area

forecasts are

highly sensitive to local migration

patterns and fertility trends (Yu et al., 2023). By using

Bayesian

hierarchical models, they are able to

“borrow strength” from neighboring areas to improve

the

robustness of projections

data-sparse regions.

Their

work demonstrates how probabilistic

models

are adaptable

to different

scales, providing a

flexible alternative to rigid deterministic structures.

Complementary to probabilistic models are

spatially explicit projections, which integrate

geographical

variation

into forecasting. Sang et

al.

developed a county-level population projection

framework in China that accounts for both spatial and

temporal dynamics using fine-grained geographic

data (Sang

et al., 2024). Their findings illustrate that

spatially-aware models can capture urbanization

trends, population clustering, and regional disparities

effectively than

conventional national-level

approaches. Chen et al. similarly adopted gridded

projections under shared socioeconomic pathways

(SSPs) for China, creating high-resolution future

population

scenarios

aligned with

global climate

frameworks (Chen et al., 2020).

Despite their strengths, probabilistic and spatial

models

face

challenges in data requirements,

computational complexity, and interpretability. Many

countries lack consistent subnational demographic

data, making localized modeling difficult. Moreover,

policymakers

may

struggle to translate probabilistic

ranges into

actionable

decisions without additional

interpretive tools.

2.3 Machine Learning and

AI-Enhanced Forecasting

Recent years have witnessed

a growing

interest in

machine learning (ML) and artificial intelligence (AI)

approaches in population forecasting, especially

where

data

complexity and

volume

exceed

the

capabilities of traditional models. One prominent

method is the Long Short-Term Memory (LSTM)

neural

network, which

excels at

modeling

temporal

dependencies in time series data. Grossman et al.

applied LSTM to small-area population forecasting in

Australia and demonstrated that these models

outperform conventional techniques

terms

predictive accuracy, particularly for short-term

forecasts (Grossman et

al., 2023).

LSTM

models

are

capable

of learning non-linear

relationships and

incorporating

lagged effects,

making them suitable for detecting sudden

demographic shifts or non-monotonic migration

flows. Their adaptability is especially useful in small-

area forecasts, where

local

trends may diverge

substantially from national averages. However,

Grossman and colleagues caution that LSTMs require

large, high-quality

training

datasets

and may be

difficult to interpret—posing barriers for their

integration into official statistical systems.

Beyond LSTMs, other AI models such as decision

trees, random forests, and

support

vector

machines

have

also been applied in demographic

contexts. For

instance, Papastefanopoulos

et al. evaluated multiple

time

series

models for forecasting COVID-19

case

proportions and demonstrated

that

machine learning

techniques, including

LSTMs and autoregressive

models, outperformed

classical

statistical models

during rapidly evolving public health scenarios

(Papastefanopoulos

al., 2020). Their work

illustrates the potential transferability of these models

to population forecasting during periods of

demographic disruption.

Despite their promise, AI models

are not without

caveats. O’Sullivan

warned

that overreliance on

opaque

models

could

exacerbate demographic

misinterpretation, particularly when policymakers

treat forecasts as deterministic outcomes.

Additionally, AI

techniques

often require extensive

computational resources and suffer from “black box”

issues, making them difficult to audit or validate

through

traditional

demographic lenses (O’Sullivan,

2023).

mitigate these

concerns, researchers

have

proposed hybrid models that

integrate traditional

demographic techniques with machine learning

enhancements. For example, baseline population

Bridging Tradition and Modernity: The Application and Reﬂection of Information Technology in Census Practices

287

projections can be generated using cohort-component

models and adjusted dynamically using LSTM

networks fed with real-time migration, birth, and

administrative data. Such

integration

improves

both

the accuracy and responsiveness of forecasts.

3 LITERATURE ANALYSIS:

APPLICATIONS AND

FORECASTING RESULTS

3.1 Datasets Used

The effectiveness

of any population forecasting

model depends significantly on the quality, resolution,

and

granularity

of the

datasets used. Across the

reviewed literature, we

observe a

diversity

data

sources

ranging from national census and

administrative

registers to

high-resolution

spatial

grids and real-time datasets.

Traditional demographic projections, as

represented by

Aryal, mainly

rely on census data,

vital registration

systems (including

birth

and death

registrations), and migration

statistics aggregated

the national

or subnational

levels. These datasets

are

typically collected every ten years and constitute

the

basis

for organized

forecasting models, such

cohort-component

projections. While these

sources

are marked by

standardization

and consistency, they

often suffer from

a lack

timeliness

and

are not

sufficient to capture rapid changes in population

dynamics, especially in times of crisis.

Probabilistic and spatial modeling approaches,

which

are represented by the research works of

Vollset

et al. and

Chen et

al. utilize

more extensive

demographic and socioeconomic information. For

example, the Global Burden of Disease study utilized

the

World Population Prospects, United

Nations

projections, and national statistical databases to make

long-term demographic projections for 195 countries.

Chen et al. extended this methodology in the Chinese

setting by integrating

gridded population

data

with

Shared Socioeconomic Pathways (SSPs) to facilitate

high-resolution

projections under alternative

climate

and

development futures. Such models

call for

harmonization of heterogeneous data sources, subject

to intricate preprocessing and spatial coordination

procedures (Vollset et al., 2020; Chen et al., 2020).

Spatial studies, such as Sang et al., relied on

county-level population data in China combined with

satellite-based urbanization metrics

and geocoded

administrative

boundaries (Sang et al., 2024). Such

fine-grained datasets are essential for modeling urban

growth and spatial population

distribution, but they

are limited by availability and standardization,

especially across developing countries.

AI and machine learning methods rely even more

heavily on comprehensive and real-time data.

Grossman et al. used small-area data from Australia’s

national

statistical

agency, which

included

historical

population counts and migration

flows

over multiple

decades (Grossman

et al., 2023). The LSTM

model

they developed required

time series input data

structured into temporal windows, making

continuous, high-frequency datasets

essential for

model

training. Similarly, Papastefanopoulos et

al.

evaluated

forecasting

models

using COVID-19

case

data expressed as percentages of active cases per

population. The

time-sensitive nature

of such data

reflects

the strengths of

AI in responding to fast-

moving demographic phenomena.

Overall, data quality and

availability remain

critical

barriers to forecasting performance. While

traditional models tolerate coarse, static data,

probabilistic and AI-based approaches demand

detailed, structured, and consistent inputs that are not

universally accessible.

3.2 Comparative Analysis of

Forecasting Results

The

reviewed

studies highlight

distinct

performance

characteristics across different forecasting techniques,

each with context-dependent strengths.

In deterministic models such as the

cohort-

component approach, Aryal found that forecasts were

generally accurate in countries with stable population

structures and low migration volatility (Aryal, 2020).

However, these

models

underperformed in

regions

experiencing sudden demographic shifts. For

example, Kim and Kim showed that even when

applied sub-nationally, deterministic projections

underestimated the speed of population

aging in

South

Korea’s rural

areas

(Kim & Kim, 2020). The

model’s simplicity, while advantageous for

interpretability, leads to

rigidity in uncertain or

rapidly evolving conditions.

Probabilistic

models introduce robustness

providing forecast intervals rather than single

estimates. Vollset et al. demonstrated that this

approach reduced errors in long-term global

projections. Their simulations, incorporating

uncertainty in fertility and migration, produced more

realistic estimates, especially for developing

countries with unstable demographic indicators

(Vollset et al., 2020). Yu et al. validated this strength

at the county level, where their Bayesian model

ICDSE 2025 - The International Conference on Data Science and Engineering

288

improved forecast precision through spatial

smoothing and uncertainty

quantification

(Yu

et al.,

2023).

Spatial models also offer improvements in

forecast precision at fine geographical scales. Sang et

al. showed that combining spatial autocorrelation

structures with temporal trend modeling allowed their

model to better

capture

migration-driven

population

clustering in

China (Sang

et al., 2024). Chen

et al. ’s

gridded

projections

further

illustrate

how the spatial

layout of population growth varies significantly under

different socioeconomic scenarios. However, these

models often require substantial computational

resources

and

geographic

data integration expertise

(Chen et

al., 2020).

Machine learning techniques have also

demonstrated

improved

short-term forecasting

small-area settings. Grossman et al. likened

predictions using LSTMs

to conventional statistical

models and

observed

that the AI

approach lowered

mean absolute percentage error (MAPE) by up to

30% for

certain regions. This

attributed

to the

model's ability to learn nonlinear patterns

and

lagged

effects over time (Grossman et al., 2023). But the

trade-off is one of

decreased

transparency—AI

predictions are frequently opaque

about causal

reasons, with ominous implications for use in policy-

sensitive areas.

Papastefanopoulos et al. also reported AI models

outperforming classical time series approaches

ARIMA in capturing the effects of the

pandemic

population movement. They, however, insisted on the

importance

of model validation and

warned

against

blind trust in accuracy metrics (Papastefanopoulos et

al., 2020).

O'Sullivan presents a critical perspective,

cautioning that excessively positive population

forecasts—regardless of whether they are founded on

conventional or

AI-driven

models—might

overlook

significant

long-term

sustainability

issues. His

argument

highlights

the

need to connect

model

outputs with inclusive policy development and

scenario evaluation (O'Sullivan, 2023).

In summary, forecasting accuracy and adaptability

vary across models:

Deterministic models

provide transparency and

simplicity but perform poorly under uncertainty.

Probabilistic and spatial models improve

reliability and geographic

resolution

but

require

complex data and calibration.

AI models deliver higher predictive accuracy,

especially short-term, but face interpretability and

data availability issues.

These trade-offs

suggest that no single method is

universally superior. Instead, hybrid strategies

combining structured demographic reasoning with

adaptive AI tools may provide the most effective path

forward.

4 RECOMMENDATIONS AND

FUTURE DIRECTIONS

The comparative analysis of existing

forecasting

methods reveals a dynamic and rapidly evolving field,

but it also underscores persistent limitations that

hinder the development

of robust, responsive, and

policy-relevant

population

projections. To

move the

field forward, several key areas warrant attention.

4.1 Addressing Data Limitations and

Standardization

A consistent issue spanning traditional, probabilistic,

and AI - based population forecasting methods is their

reliance on high - quality, uniform data. Aryal points

out that even the well - established cohort -

component models encounter difficulties when

dealing with incomplete vital statistics or inconsistent

census

intervals

(Aryal ，2020). This

problem is far

more acute in machine learning models. For instance,

as Grossman et

al. show, in such models, real - time

and temporally consistent input data

are essential for

model training and validation (Grossman et al., 2023).

To foster progress in these methodologies,

national statistical

agencies and international bodies

should make data modernization a priority. This

involves

digitizing records, enhancing population

estimates between

censuses, and implementing open

data standards. The effectiveness of probabilistic

models, such as those developed by Vollset et al. and

Yu et al., hinges largely on having detailed and

consistent input variables across

various regions and

time periods (Vollset et al., 2020; Yu et al., 2023).

Moreover, global investments in high - resolution

spatial

data, similar to

those used in the studies by

Chen

et al. and

Sang et al., can enhance

small -

area

projections and allow for integration with

environmental, economic, and urban planning models.

The future

of population forecasting lies

in the real

time integration of data

from multiple sources

(Chen

et al., 2020; Sang et al., 2024).

Bridging Tradition and Modernity: The Application and Reﬂection of Information Technology in Census Practices

289

4.2 Enhancing Model Interpretability

and Transparency

AI and machine

learning

models

have

shown great

potential in improving short-term forecasting

accuracy, especially in small-area contexts

(Grossman et al., 2023; Papastefanopoulos et al.,

2020). However, their

“black-box” nature often

makes them unsuitable for high-stakes policy

environments, where

decision-makers require

clear,

interpretable evidence for interventions.

Developing explainable AI (XAI) frameworks

tailored

to demographic forecasting is

therefore

essential. These could involve hybrid approaches,

where

traditional

demographic

structures are used to

anchor AI models, offering both predictive power and

theoretical transparency. For

example, using LSTM-

based adjustments on top of deterministic cohort

estimates may allow practitioners to preserve

interpretability while enhancing responsiveness to

new data

inputs.

Moreover, as O’Sullivan cautions,

overconfidence in model precision—whether from

classical or AI-based methods—can distort policy

planning and delay recognition of demographic risks.

Researchers

should

routinely publish

uncertainty

estimates, model assumptions, and validation metrics

alongside forecasts (O’Sullivan, 2023).

4.3 Bridging Scale and

Scenario Gaps

Another persistent challenge is the mismatch between

global or national forecasting and the needs of local

policymakers. As demonstrate in their study of South

Korea, national trends often conceal sharp

subnational divergence, particularly in aging,

urbanization, and

fertility. To address this, models

must become more spatially adaptive, integrating

geographic heterogeneity and local socioeconomic

indicators.

Scenario-based forecasting also deserves more

attention. While Chen

et al. adopted Shared

Socioeconomic Pathways (SSPs) in population

modeling

for

China, few national statistical agencies

implement scenario planning into their population

projections. Probabilistic methods and spatial models

are particularly well-suited to

scenario modeling,

offering

a pathway to

incorporate uncertainty

fertility

preferences, climate impact, and migration

policy shifts (Chen et al., 2020).

Developing standardized scenario frameworks,

similar to those in climate

modeling, could

significantly enhance the robustness and relevance of

demographic forecasts.

4.4 Strengthening Interdisciplinary

Integration

Demographic

forecasting

must

increasingly operate

at the intersection

of public

health, climate science,

urban

planning, and

artificial intelligence. Vollset

al. provide an excellent model by integrating

population projections into health burden

forecasting

(Vollset

al., 2020), while Sang

et al. show

how

urban expansion models can inform localized

demographic dynamics (Sang et al., 2024).

Future forecasting frameworks should support

plug-and-play

interoperability

with

other

models—

such

those used

epidemiology, infrastructure

planning, and environmental simulation. This

requires

not only

methodological compatibility but

also institutional collaboration and shared data

platforms.

4.5 Supporting Global Equity in

Forecasting Capacity

Many of the advanced models and datasets examined

this paper

originate from

high

income

countries

boasting well - established statistical systems.

However, as global demographic trends

increasingly

center on

low

and middle - income regions, it is

crucial to

enhance

forecasting

capabilities

in areas

with limited

data.

Efforts to create open

- source

modeling tools,

implement

capacity - building programs for national

statistics offices, and establish collaborative

international

datasets (such as those

the

Global

Burden of

Disease study) are essential. These

initiatives ensure that all countries, regardless of their

income level or data availability, can generate

population forecasts that are both reliable and

relevant to policy - making.

5 CONCLUSIONS

Population forecasting is vital for planning

infrastructure, healthcare, labor

markets, and social

services. In a world

of demographic uncertainty and

rapid

global change, accurate, adaptable, and

clear

forecasting tools are urgently needed. This paper

compares three main population forecasting methods:

traditional demographic models, probabilistic and

spatial models, and machine - learning

- based ones.

Each method

has its

own strengths, depending

application scale, data availability, and population

trend volatility.

ICDSE 2025 - The International Conference on Data Science and Engineering

290

Traditional cohort - component models are

fundamental as

they are

interpretable

and

work well

with

census

data. However, they often

can't

capture

sudden demographic shifts, like those from

unexpected migrations. Also, they struggle with local

- level variations where

regional

differences in

economic or cultural factors affect population trends.

Probabilistic approaches, such as those by Vollset

et al. and Yu et al., address these issues. By factoring

in uncertainty, they provide a range of possible future

population scenarios, thus enhancing forecast

reliability, especially at national

and

regional

levels.

Spatial forecasting and gridded projections enable

more detailed

demographic planning, crucial for

rapidly urbanizing, diverse countries. Machine

learning techniques

like LSTM networks are good at

catching non

linear trends

for

small -

area

short -

term forecasts. But, as Papastefanopoulos

al. and

O’Sullivan

note, they have

challenges

in data

needs

and interpretability.

Integrating these methods

holds promise. Hybrid

models combining

demographic

theory and AI can

offer

machine - learning accuracy and

traditional -

model

transparency. Improving data infrastructure,

investing in open - access tools, and developing

explainable AI will boost forecasting systems. In

essence, refining

population

forecasting is

societal

must

for

creating resilient, inclusive 21st -

century

policies as demographic shifts shape our future.

REFERENCES

Aryal, G. R. 2020. Methods of Population

Estimation and

Projection. Journal of

Population and Development,

1(1), 54–61.

Chen, Y., Guo, F., Wang, J., &

al. 2020. Provincial and

gridded

population

projection for

China under

shared

socioeconomic pathways from 2010 to 2100. Scientific

Data, 7, 83.

Ellner, S. P., Adler, P. B., Childs, D. Z., Hooker, G., Miller,

T. E. X., &

Rees, M. 2022. A critical comparison of

integral

projection and matrix

projection models for

demographic analysis: Comment. Ecology, 103(10), 1–

Grossman, I., Wilson, T., & Temple, J. 2023. Forecasting

small area

populations with long

short-term

memory

networks. Socio-Economic Planning Sciences, 88,

101658.

Kim, K. W., & Kim, O. S. 2020. Super aging in

South

Korea unstoppable but mitigatable: A sub-national

scale

population projection for

best

policy planning.

Spatial Demography, 8(2), 155–173.

O’Sullivan, J. N. 2023. Demographic

Delusions: World

Population

Growth Is Exceeding Most Projections and

Jeopardising Scenarios for Sustainable Futures. World,

4(3), 545-568.

Papastefanopoulos, V., Linardatos, P., & Kotsiantis, S.

2020. COVID-19:

A Comparison of Time Series

Methods

to Forecast Percentage of Active

Cases per

Population. Applied Sciences, 10(11), 3880.

Sang, M., Jiang, J., Huang, X., &

al. 2024. Spatial and

temporal changes in population distribution and

population projection at county level in China.

Humanities

and

Social Sciences

Communications, 11,

288.

Vollset, S. E., Goren, E., Yuan et al., 2020. Fertility,

mortality, migration, and population scenarios for 195

countries and territories from 2017 to 2100: A

forecasting analysis

for

the Global

Burden of Disease

Study. The Lancet, 396(10258), 1285–1306.

Yu, C. C., Ševčíková, H., Raftery, A. E., & Curran, S. R.

2023. Probabilistic county-level population projections.

Demography, 60(3), 915–937.

Bridging Tradition and Modernity: The Application and Reﬂection of Information Technology in Census Practices

291