upload_ykut55hfo6at6ed7bu4wt4wjbx.pdf

72

│

R or QGIS? A Case Study on Raster Analysis

Efficiency in Agricultural Engineering Education

¿R o QGIS? Un estudio de caso sobre la eficiencia

del análisis de imágenes ráster en la enseñanza

de ingeniería agronómica

Abstract

Geographic Information Systems (GIS) are essential in agricultural and environmental sciences, yet handling complex spatial data requires ecient,

reproducible workows. This study evaluated the operational eciency and student perception of R (via R Markdown) versus QGIS for raster analy

-

sis. Fifty-two agronomic engineering students from the University of Cuenca completed a raster temperature extraction exercise using both platforms.

Survey responses and completion times were analyzed using McNemar’s tests, exact binomial tests, and generalized linear models. Results showed

that exercise completion was signicantly higher in R (98%) than in QGIS (87%; P=0.031). R demonstrated superior time eciency, with 65% of stu

-

dents nishing within three minutes compared to only 29% in QGIS. Additionally, students reported signicantly greater self-perceived understan

-

ding of the workow when using R (71% vs. 29%; P<0.01). While preference for a primary GIS teaching platform was evenly split (QGIS 52%, R 48%;

P>0.05), an overwhelming majority (96%; P<0.001) indicated they would adopt R as a complementary tool due to execution speed and workow

transparency. Although instructional format asymmetry presents a potential confounding factor, these ndings demonstrate that integrating repro

-

ducible scripting environments into GIS education enhances computational eciency, user condence, and procedural clarity in raster processing.

Keywords:

Rstudio, Rmarkdown, GIS, spatial analysis, learning, raster

Resumen

Los Sistemas de Información Geográca (SIG) son esenciales en las ciencias agrícolas y ambientales; sin embargo, el manejo de datos espaciales com

-

plejos requiere ujos de trabajo ecientes y reproducibles. Este estudio evaluó la eciencia operativa y la percepción estudiantil de R (mediante R Mark

-

down) frente a QGIS para el análisis ráster. Cincuenta y dos estudiantes de ingeniería agronómica de la Universidad de Cuenca completaron un ejercicio

de extracción de temperatura ráster en ambas plataformas. Las respuestas de la encuesta y los tiempos de ejecución se analizaron mediante la prueba de

McNemar, pruebas binomiales exactas y modelos lineales generalizados. Los resultados mostraron que la nalización del ejercicio fue signicativamente

mayor en R (98%) que en QGIS (87%; P = 0.031). R demostró mayor eciencia temporal: el 65% de los estudiantes terminó en tres minutos o menos,

frente a solo un 29% en QGIS. Además, los participantes reportaron una comprensión autopercibida del ujo de trabajo signicativamente mayor al

usar R (71% frente a 29%; P < 0.01). Aunque la preferencia por la plataforma principal de enseñanza estuvo dividida (QGIS 52%, R 48%; P > 0.05), una

abrumadora mayoría (96%; P < 0.001) indicó que adoptaría R como herramienta complementaria debido a su rapidez de ejecución y transparencia me

-

todológica. Aunque la asimetría del formato instruccional representa un potencial factor de confusión, estos hallazgos demuestran que integrar entornos

de programación reproducibles en la educación SIG fortalece la eciencia computacional, la conanza del usuario y la claridad procedimental.

Palabras clave:

Rstudio, Rmarkdown, SIG, análisis espacial, aprendizaje, ráster

Recibido:

05 de mayo de 2026

Aceptado:

24 de agosto de 2026

Alberto Macancela-Herrera

1*

; Eduardo Tacuri-Espinoza

2

;

Lucía Lupercio-Novillo

3

; Mateo López-Espinoza

4

1

Laboratorio de Geomática, Facultad de Ciencias Agropecuarias, Universidad de Cuenca, Cuenca, Ecuador; GeoinfoRmática, Agrociencias

y

Soluciones

Sostenibles

(G.R.A.S.S);

Universidad

de

Cuenca,

Ecuador;

alberto.macancelah@ucuenca.edu.ec;

https://orcid.

org/0000-0003-1461-4364

2

Laboratorio de Geomática, Facultad de Ciencias Agropecuarias, Universidad de Cuenca, Cuenca, Ecuador; GeoinfoRmática, Agrociencias

y Soluciones Sostenibles (G.R.A.S.S); Universidad de Cuenca, Cuenca, Ecuador; eduardo.tacuri@ucuenca.edu.ec; https://orcid.org/0000-

0002-4094-209X

3

Laboratorio de Geomática, Facultad de Ciencias Agropecuarias, Universidad de Cuenca, Cuenca, Ecuador; GeoinfoRmática, Agrociencias

y Soluciones Sostenibles (G.R.A.S.S), Universidad de Cuenca, Cuenca, Ecuador; lucia.lupercio@ucuenca.edu.ec; https://orcid.org/0000-

0002-4798-6108

4

Laboratorio de Geomática, Facultad de Ciencias Agropecuarias, Universidad de Cuenca, Cuenca, Ecuador; GeoinfoRmática, Agrociencias

y Soluciones Sostenibles (G.R.A.S.S), Universidad de Cuenca, Cuenca, Ecuador; mateo.lopeze@ucuenca.edu.ec; https://orcid.org/0000-

0002-6996-8701

*

Autor de correspondencia

Revista Ciencia UNEMI

Vol. 19, N° 52, Septiembre-Diciembre 2026

, pp. 72 - 83

ISSN 1390-4272 Impreso

ISSN 2528-7737 Electrónico

https://doi.org/10.29076/issn.2528-7737vol19iss52.2026pp72-83p

│

73

Macancela

.

R or QGIS? A Case Study on Raster Analysis Eciency

I. INTRODUCTION

A

Geographic

Information

System

(GIS)

comprises

integrated

hardware

and

software

for

storing,

analyzing

and

displaying

geographically

referenced

data.

These

systems

enable

the

identication

of

spatial

patterns,

trends

and

relationships

through

validated

spatial

analysis

(Scholten,

1990).

GIS

has

become

increasingly

relevant in agriculture and environmental sciences,

where advanced geospatial techniques are employed

to

monitor

crop

performance,

assess

vegetation

dynamics

and

evaluate

ecological

parameters.

Beyond

scientic

applications,

GIS

methodologies

have

also

been

adopted

in

marketing

and

business

analytics to optimize spatial decision making (Turk

et al., 2014; Li et al., 2024).

Specialized

programs

now

include

tools

to

download,

visualize,

analyze

and

graph

spatial

information

(Zhou,

2025).

Among

these,

QGIS

(Quantum

GIS)

is

a

widely

used

open-source

desktop

GIS.

It

benets

from

a

global

community

that

continuously

contributes

to

its

development,

documentation and technical support. Its integrated

interface

enhances

usability,

making

it

accessible

to

both

novice

and

advanced

users.

The

extensive

collection of functions, tools and plugins makes QGIS

a powerful platform (Li et al., 2024). Nevertheless,

several

users

have

reported

problems

with

layer

alignment,

software

performance

and

interface

glitches,

which

can

be

counterproductive

when

managing large raster datasets (Wu et al., 2024).

The incorporation of programming languages into

GIS education has become increasingly important as

spatial datasets continue to grow in size, complexity

and

diversity.

Traditional

graphical

GIS

interfaces

are highly valuable for visualization and exploratory

analysis; however, they may become inecient when

repetitive

operations,

large

raster

collections

or

automated

workows

are

required.

In

this

context,

learning programming languages such as R provides

students and professionals with the ability to create

reproducible,

transparent

and

scalable

geospatial

analyses

(Pebesma

and

Bivand,

2023).

Through

scripting, users can document every analytical step,

reducing human error and facilitating replication of

results by other researchers.

R is a high-level programming language originally

developed

for

statistical

analysis.

Its

application

in

GIS and spatial analysis has expanded considerably

due to the development of specialized packages that

integrate geospatial functionalities (Saqr and López

Pernas,

2024).

R

is

now

regarded

as

one

of

the

most important tools for statistical and spatial data

analysis

and

has

become

essential

in

research

and

general data processing workows (Parra et al., 2023;

Graser et al., 2025; Tucker et al., 2023). Compared

with

integrated

GIS

platforms

such

as

QGIS

or

ArcGIS

Pro,

R

often

requires

fewer

computational

resources, making it a more ecient alternative for

large

scale

raster

processing

and

spatial

modelling

(Nguyen Tien et al., 2019). In addition, R integrates

data

acquisition,

preprocessing,

spatial

analysis,

statistical modelling and visualization within a single

environment.

Packages

such

as

terra,

sf,

tidyverse

and

ggplot2

allow

ecient

manipulation

of

raster

and

vector

data

while

maintaining

compatibility

with

modern

data

science

workows

(Hijmans,

2025;

Pebesma,

2018;

Wickham

et

al.,

2019).

This

integration

is

particularly

advantageous

in

agricultural

engineering,

where

spatial

information

is

frequently

combined

with

experimental,

climatic

and socioeconomic datasets. Furthermore, learning

R

promotes

coding

skills

that

can

be

transferred

to

other

programming

environments,

thereby

strengthening

students'

computational

thinking,

analytical reasoning and problem-solving abilities in

GIS applications.

Raster

data

are

digital

images

composed

of

grids

of

cells

or

pixels

that

collectively

form

an

overall

image.

Raster

resolution

depends

on

pixel

size; therefore, the smaller the pixel, the higher the

resolution. Each pixel possesses a value determined

by the information contained in the image (Poynton,

2003).

Contemporary

raster

datasets

originate

from

diverse

sources,

including

satellite

imagery,

orthophotographs, unmanned aerial vehicles (UAVs)

and

digital

elevation

models

(DEMs),

all

of

which

are

fundamental

inputs

for

geospatial

analysis

and

environmental monitoring (Singla et al., 2021). The

use

of

raster

data

in

agriculture

and

ecology

has

increased

because

of

the

wide

availability

of

raster

sources

(Steiniger

et

al.,

2017).

One

specic

application

is the calculation and analysis of vegetation indices,

which help improve crop management by identifying

stress,

disease

or

nutrient

deciency

(Rasul

et

al.,

2020). Another important use is in climate studies,

74

│

Volumen 19, Número 52, Septiembre-Diciembre 2026, pp. 72 - 83

where

raster

data

model

climatic

conditions

for

agriculture, biodiversity and climate change research

(Martin

Gomez

and

Bartolome

Muñoz

de

Luna,

2023).

From

an

educational

perspective,

the

use

of

programming

languages

encourages

active

learning.

Instead

of

following

only

predened

graphical

procedures,

students

must

understand

the

logical

sequence

of

the

analysis,

the

structure

of

spatial

objects

and

the

relationships

between

inputs

and

outputs.

have

shown

that

programming-based

instruction

can

improve

analytical

reasoning,

autonomy

and

long-term

retention of methodological concepts (Tucker et al.,

2023).

In

GIS,

this

means

that

students

not

only

learn how to generate maps but also how to design

workows capable of processing hundreds of spatial

layers

automatically.

Another

relevant

advantage

is

that

R

is

free

and

open

source,

which

facilitates

its

adoption

in

universities

with

limited

economic

resources and promotes equitable access to advanced

geospatial

tools.

Consequently,

integrating

R

and

other programming languages into GIS curricula may

strengthen both technical competencies and research

capacity

in

future

agricultural

and

environmental

professionals.

In this context, R oers the facility to download

raster data from multiple repositories using, in most

cases, only one line of code (Giofandi et al., 2023). We

hypothesized that the R Markdown workow would

enable

students

to

complete

the

raster

extraction

exercise

more

eciently

(higher

completion

rates

and

shorter

times)

and

with

greater

perceived

understanding

than

the

QGIS

handbook-based

workow. We interpret eciency and comprehension

as outcomes associated with the scripted, scaolded

nature of the R environment, rather than as inherent

properties

of

the

R

software

itself.

Based

on

our

ndings, we propose that incorporating R and similar

scripted

tools

as

complementary

resources

in

GIS

curricula can enhance reproducibility and analytical

clarity,

while

supplementary

coding

instruction

is

needed to address syntax-related challenges.

II. MATERIALS AND METHODS

Data Collection

Students

of

agronomical

engineering

at

the

University of Cuenca performed the exercise in QGIS

and R programs. Subsequently, a questionnaire was

administered

during

August

2025.

Participation

required

students

to

meet

three

criteria:

(i)

completion or current enrollment in courses related

to

GIS,

topography,

or

precision

farming,

and

(ii)

prior

experience

with

at

least

one

geospatial

software package (QGIS, ArcMap, ArcGIS Pro) and

(iii)

understand

the

logic

behind

code

generation

to

R

programming

language.

A

methodological

handbook for QGIS was developed to guide students

through the exercise. In parallel, an R Markdown was

developed

to

replicate

the

same

process,

detailing

each

line

of

code.

This

approach

was

intended

to

help students associate new GIS code with statistical

code learned previously (statistics and experimental

design). We received survey results from 52 students

to

perform

results.

The

order

in

which

students

performed

the

exercises

was

not

predetermined;

each

participant

independently

decided

whether

to

begin with R or QGIS. Consequently, the sequence of

execution varied across students and was dependent

on individual choice.

R and QGIS exercise

R and QGIS were software used to carry out the

extraction of pixel value from raster. It was necessary

to

use

a

raster

temperature

average

image

from

Ecuador, which was downloaded from the database

WorldClim

with

0.5

min

arc

pixel

size

(~

1

km).

In

general,

the

exercise

consisted

of:

1)

download

raster

data,

2)

save

raster

data

(step

1

and

2

were

not evaluated for the questionnaire), 3) open raster

le, 4) visualize raster, 5) georeferenced, 6) extract

raster

values

using

points

from

a

data

frame,

7)

create a database in .csv le. For QGIS exercise, the

raster

was

downloaded

directly

from

WorldClim

(www.worldclim.org)

using

a

search

engine.

The

remaining procedure basic function from QGIS was

executed.

On

the

other

hand,

for

R

program

the

raster download was carried out in the same script,

although, last process mentioned was not evaluated.

The students must identify, understand and execute

every step for both programs. We recommend using

the last version of RStudio and QGIS LTR 3.40.

Exercise survey

The questionnaire elaborated to validate the use

of R included 8 questions. All these have a relation

│

75

Macancela

.

R or QGIS? A Case Study on Raster Analysis Eciency

with

QGIS

and

R

programs

because

of

the

raster

exercise.

With

the

results

of

the

questionnaire,

we

searched for validating students' use of R. Below are

the questions that were part of the survey:

1.

Are you familiar with the software?

2.

Were you able to complete the exercise?

3.

Was

it

complicated

for

you

to

use

the

software?

4.

If

your

last

answer

was

yes,

what

is

the

reason?

5.

How

long

did

it

take

you

to

complete

the

exercise?

6.

In

which

software

did

you

best

understand

the procedure to extract raster image data?

7.

What

software

would

you

use

as

the

main

tool for GIS?

8.

Would you use R as complementary software

and why?

Data analysis

A

survey

consisting

of

eight

questions

was

designed through Google Forms. From the analysis

of

the

questionnaire

results,

we

calculated

the

percentages

for

each

question.

Of

the

eight

survey

questions, only Questions 6 and 7 were subjected to

inferential analysis using generalized linear models

(GLMs) with a binomial distribution, because these

questions directly addressed the primary comparative

objectives

of

the

study:

students'

perceived

understanding

of

the

raster

extraction

procedure

and

their

preferred

software

for

GIS

applications.

The remaining questions (1–5 and 8) were analyzed

descriptively

because

they

were

intended

primarily

to characterize previous familiarity, task completion,

perceived diculty, completion time, and willingness

to use R as a complementary tool. Thus, percentages

were considered sucient to describe these aspects

of

the

participants’

experience,

whereas

Questions

6

and

7

allowed

direct

comparison

of

categorical

responses between R and QGIS. xercise completion

rates

(Question

2)

were

additionally

compared

between R and QGIS using McNemar's test for paired

binary outcomes, given that each student completed

the

exercise

under

both

conditions.

Willingness

to

adopt R as a complementary tool (Question 8) was

evaluated

against

a

null

expectation

of

50%

using

an

exact

binomial

test.

In

addition,

bar

plots

were

generated to visually represent the percentage-based

responses. All data management and graphics were

obtained through R with the ggplot2 library (R Core

Team, 2026).

III. RESULTS

According

to

the

initial

survey

question,

all

students

reported

familiarity

with

QGIS,

whereas

only

two

respondents

indicated

that

they

were

not

familiar

with

R

(Figure

1a).

After

the

execution

of

the

exercise,

we

found

an

interesting

result:

98%

of

participants

successfully

completed

the

raster

extraction

exercise

in

R

using

the

R

Markdown

script,

a

percentage

signicantly

higher

than

the

87% obtained for QGIS (McNemar's test, P = 0.031;

Figure 1b). Likewise, 15% of the students indicated

that

they

experienced

diculties

completing

the

extraction

of

raster

values

with

both

software

packages

(Figure

1c).

The

principal

reasons

are

presented in Figure 1d. For QGIS, the main limitation

was

the

graphical

interface,

followed

by

diculties associated with the coordinate reference

system (CRS), software lag, and the identication of

appropriate tools for the procedure. In contrast, for

R,

the

reported

diculties

were

mainly

associated

with coding syntax and compatibility issues related

to

the

RStudio

version.

Despite

these

challenges,

the

higher

completion

rate

observed

in

R

suggests

that the scripted and sequential workow provided

by R Markdown facilitated a more standardized and

reproducible execution of the raster analysis among

participants.

76

│

Volumen 19, Número 52, Septiembre-Diciembre 2026, pp. 72 - 83

Figure 1.

Software familiarity, exercise completion, perceived difficulty, and reported

operational obstacles for R and QGIS (n = 52). a) Prior software familiarity. b) Workflow

completion rates. c) Perceived operational difficulty. d) Categorization of primary

obstacles encountered in each platform.

Figure 2.

Operational efficiency and self-perceived understanding of raster processing in R and

QGIS (n = 52). a) Distribution of exercise completion times across four time intervals

(1, 3, 5, and >5 min). b) Proportion of students reporting higher self-perceived

comprehension of the analytical workflow by software platform (P < 0.01).

The

time

required

to

complete

the

raster

extraction exercise diered between the two software

environments

and

was

lower

when

using

the

R

Markdown

script

than

when

using

QGIS

(Figure

2a).

For

the

R/R

Markdown

workow,

15%

of

students

completed

the

exercise

in

approximately

one

minute,

50%

in

approximately

three

minutes,

31% in approximately ve minutes, and 4% required

more than ve minutes. In contrast, students using

QGIS generally required approximately ve minutes

or

more

to

complete

the

exercise,

with

a

greater

proportion

of

participants

requiring

more

than

ve minutes compared with the R workow. These

results

indicate

that

the

R

was

associated

with

shorter

and

more

homogeneous

completion

times

Responses

to

Question

7

(Figure

3a)

revealed

that

the

comparison

between

QGIS

and

R

was

not

statistically

signicant

(P

>

0.05),

indicating

that

both

software

packages

showed

similar

levels

of

acceptance

among

the

evaluated

students.

Specically,

52%

of

participants

preferred

QGIS

as

the primary software for GIS classes, whereas 48%

selected

R.

This

result

is

particularly

noteworthy

considering

that,

for

most

students,

this

was

the

under the instructional conditions evaluated in this

study. However, this dierence should be interpreted

considering

the

dierent

levels

of

instructional

guidance

provided

for

each

software

environment.

Furthermore,

Figure

2b

shows

that

a

signicantly

greater

proportion

of

students

reported

a

better

understanding

of

the

extraction

process

when

using

R

than

when

using

QGIS

(P

<

0.01).

These

ndings suggest that the sequential and reproducible

structure

provided

by

R

Markdown

was

associated

with lower execution times and higher self-reported

understanding

of

the

analytical

workow

and

the

relationship

between

each

processing

step

and

the

nal raster output.

rst experience using R for geospatial applications,

suggesting a rapid adaptation to the programming-

based workow. The relatively balanced distribution

of preferences indicates that students recognized the

potential of R despite their limited prior exposure to

spatial coding.

According

to

the

nal

survey

question,

most

respondents (96%) stated that they would use R as a

complementary tool for GIS analysis (exact binomial

│

77

Macancela

.

R or QGIS? A Case Study on Raster Analysis Eciency

test, P < 0.001). The principal reason was the rapid

execution

of

code,

which

represented

78%

of

the

responses,

followed

by

the

simplicity

of

generating

plots and presenting results (10%), and the greater

clarity in understanding the analytical process (6%).

Additional comments indicated that students valued

the

possibility

of

reproducing

the

entire

workow

and

modifying

the

analysis

with

minimal

manual

intervention.

In

contrast,

the

small

proportion

In

addition

to

the

completion

time,

the

distribution

of

responses

suggests

a

more

homogeneous

performance

among

students

when

using

R.

Most

participants

completed

the

exercise

within

a

narrow

time

range,

indicating

lower

variability in execution compared with QGIS, where

completion times were more dispersed. This pattern

may

reect

the

standardized

workow

provided

by

the

R

Markdown

script,

which

guided

students

through

each

analytical

step

in

a

sequential

and

reproducible

manner.

Furthermore,

the

higher

success rate observed in R was achieved even though

most students had no previous experience applying

R to geospatial analyses, suggesting that the scripted

environment reduced operational uncertainty during

raster

extraction.

The

questionnaire

also

revealed

that diculties associated with QGIS were primarily

interface

management

and

coordinate

reference

system

handling,

whereas

challenges

in

R

were

mainly

associated

with

coding

syntax

and

software

version

compatibility.

These

results

indicate

that

the

obstacles

encountered

in

QGIS

were

predominantly

procedural,

while

those

in

R

were

technical

and

potentially

easier

to

overcome

through

additional

programming

practice.

Overall,

the

ndings

support

the

interpretation

that

R

not

only improved eciency but also promoted a more

consistent and reproducible execution of the raster

analysis workow among participants.

of

students

who

were

not

inclined

to

adopt

R

as

complementary

software

mainly

attributed

their

decision

to

diculties

coding

logic,

syntax,

and

limited

experience

with

programming

environments

(Figure

3b).

These

results

suggest

that,

although

QGIS

remains

slightly

preferred

as

the

principal

teaching

platform,

students

perceive

substantial advantages in incorporating R into GIS-

based analytical workows.

IV. DISCUSSION

A critical methodological consideration must be

addressed

upfront

when

interpreting

the

ndings

of this study. The comparison between R and QGIS

was

confounded

by

a

structural

asymmetry

in

the

instructional

materials

provided

for

each

software

environment.

Specically,

the

R

Markdown

script

guided

students

through

each

analytical

step

in

a

fully

sequential,

executable

format,

whereas

the

QGIS

handbook

provided

written

instructions

that

required

students

to

independently

locate

and

execute each tool within the graphical interface. This

asymmetry

in

the

level

of

instructional

scaolding

likely

contributed

to

the

higher

completion

rates,

shorter

execution

times,

and

greater

self-perceived

understanding

observed

in

the

R

workow,

independent

of

any

inherent

advantage

of

the

software itself.

Therefore,

we

frame

our

conclusions

not

in

terms

of

R

versus

QGIS

as

software

platforms,

but

rather in terms of scripted, reproducible workows

versus

manual

GUI-based

workows.

The

scripted

environment

reduced

operational

uncertainty,

provided

immediate

feedback,

and

explicitly

documented

each

analytical

step

pedagogical

features

that

can

improve

student

performance

regardless

of

the

underlying

software.

This

distinction is essential for correctly interpreting our

results and for designing future comparative studies.

Figure 3.

Student platform selection for primary GIS instruction and complementary tool

adoption. a) Preferred primary GIS software selection. b) Willingness to adopt R as a

complementary GIS tool partitioned by underlying motivations.

78

│

Volumen 19, Número 52, Septiembre-Diciembre 2026, pp. 72 - 83

Future

research

should

consider

standardizing

the

instructional

format

across

platforms

(e.g.,

equally

detailed step-by-step guides for both QGIS and R) or

using a counterbalanced design to isolate the eect of

the software from the eect of the guidance material.

Previous research

Bearing

this

limitation

in

mind,

our

ndings

indicate

that

the

R

Markdown

workow

was

associated

with

greater

eciency

than

the

QGIS

handbook-based workow for extracting pixel values

in basic raster exercises. This supports the growing

body of literature that highlights the advantages of

programming-based

workows

for

spatial

analysis

(Pebesma and Bivand, 2023). The higher completion

rate

and

shorter

execution

time

observed

in

our

study suggest that scripted environments can reduce

operational

complexity

and

improve

consistency

during

geospatial

procedures.

In

addition,

the

use

of R Markdown enhanced students' comprehension

by

providing

a

transparent

and

reproducible

workow in which every analytical step was explicitly

documented.

Reproducibility

is

increasingly

recognized

as

a

fundamental principle in spatial and environmental

research, and scripting languages such as R facilitate

the

replication,

verication,

and

extension

of

analyses (Pebesma and Bivand, 2023). The ability to

download, preprocess, analyze, and visualize raster

data within a single computational environment also

represents

a

substantial

advantage

for

agricultural

and

environmental

applications,

where

large

datasets

are

frequently

used

(Nguyen

Tien

et

al.,

2019). Previous studies have recommended the use

of

programming

languages

for

handling

extensive

raster information because they allow automation of

repetitive tasks and reduce the limitations associated

with

manual

graphical

workows

(Giofandi

et

al.,

2023; Zhou et al., 2018).

Our results are consistent with previous studies

reporting

that

programming-based

approaches

provide

greater

eciency

and

reproducibility

for

spatial

analysis

than

conventional

graphical

GIS

workows. Nguyen Tien et al. (2019) demonstrated

that

R

oers

high

versatility

for

processing

large

geospatial

datasets,

particularly

in

environmental

and

agricultural

applications.

Similarly,

Pebesma

and Bivand (2023) emphasised that scripting-based

workows

improve

transparency,

scalability,

and

reproducibility compared with manual operations in

desktop GIS software. In educational contexts, Tucker

et

al.

(2023)

showed

that

learning

R

strengthens

analytical reasoning and data management skills.

Moreover, the availability of specialized packages

has

considerably

expanded

the

capacity

of

R

for

ecient

raster

and

vector

processing.

Although

none of the participants had prior experience using

R

for

GIS

applications,

nearly

half

selected

it

as

their

preferred

tool,

suggesting

that

well-designed

coding

environments

may

facilitate

the

adoption

of

programming

approaches

even

among

novice

geospatial users. R is a exible and powerful statistical

software environment widely applied to GIS analysis,

and it has been increasingly adopted by researchers,

professionals, and students across the social sciences,

humanities,

and

STEM

disciplines

(Tucker

et

al.,

2023). R has been specialized in spatial analysis due

to the development of libraries (e.g., raster or terra),

facilitating the operation of raster images (Hijmans,

2020). Several important research projects have used

R and these libraries to obtain results for agriculture,

climate,

and

vegetation

health

(Nguyen

Tien

et

al.,

2019).

In

Peru,

the

introduction

of

programming

languages

in

public

universities

revealed

eciency

for topics that involve mathematics (Paucar-Curasma

et al., 2023). According to our survey results, the R

Markdown

workow

was

associated

with

shorter

execution

times,

and

participants

reported

higher

self-perceived

comprehension

of

the

extraction

procedure compared with QGIS.

Broader

Educational

and

Research

Context

Earlier

research

has

shown

that

students

who

engage with programming languages achieve higher

academic

performance

in

lectures,

largely

due

to

their

enhanced

problem-solving

capacity

and

skill

development

(Cuevas

et

al.,

2025).

This

nding

is

particularly

relevant

in

the

context

of

our

study,

where

the

scripted

R

workow

appears

to

have

facilitated not only task completion but also a clearer

understanding of the analytical process. Furthermore,

from 1996 to 2022, scientic production in Europe

increased considerably; this phenomenon responds

mainly to reforms that occurred 20 years ago within

the higher education system of the European Union

│

79

Macancela

.

R or QGIS? A Case Study on Raster Analysis Eciency

(

K

n

u

t

a

s

et

al.,

2021).

Notably,

a

key

nding

of

that

trend

was the higher rate of scientic publication among

researchers

who

developed

the

capability

to

utilize

a programming language. In this regard, our study

specically

highlights

the

use

of

R

(Martin

Gomez

and

Bartolome

Muñoz

de

Luna,

2023),

suggesting

that

this

programming

language

has

the

potential

to

develop

inherent

abilities

for

future

research

among

students

who

employ

R

as

a

resource

for

data

and

spatial

analysis.

Thus,

integrating

R

into

GIS

education

may

not

only

improve

immediate

task performance but also foster long-term research

capacities and scholarly output.

Pedagogical

Challenges

and

Implications

for GIS Education

Despite

the

growing

relevance

of

programming

languages

in

geospatial

sciences,

the

incorporation

of

R

into

GIS

education

continues

to

present

important

pedagogical

challenges.

have

reported

that

students

frequently

experience

diculties

when

transitioning

from

graphical

interfaces

to

code-based

analytical

environments,

particularly

because

programming

requires

the

simultaneous

understanding

of

syntax,

logical

structures, data organization, and analytical reasoning

(Pavlenko et al., 2022). These cognitive demands are

often intensied in spatial analysis, where students

must

also

comprehend

coordinate

systems,

raster

and vector structures, and the sequential nature of

geoprocessing workows. Consequently, high levels

of

frustration,

reduced

self-condence,

and

lower

persistence

have

been

documented

in

introductory

programming

courses

(Mladenović

et

al.,

2017).

Educational research suggests that these diculties

can

be

mitigated

through

scaolded

instruction,

guided

practice,

and

interactive

learning

activities

that

progressively

connect

theoretical

concepts

with

practical

applications

(Portella-Cleves

and

Rodríguez-Hernández, 2024).

In

our

study,

the

use

of

an

R

Markdown

script

functioned

as

a

structured

learning

support,

allowing

students

to

visualize

each

command

and

immediately

observe

its

eect

on

the

raster

analysis.

This

step-by-step

approach

likely

reduced

cognitive

overload

and

facilitated

the

comprehension of geospatial procedures, reinforcing

the value of reproducible and guided programming

environments

in

GIS

education.

Furthermore,

the

challenges

identied

by

participants

regarding

R

syntax and coding should not be interpreted solely as

limitations of the software, but rather as indicators

of

insucient

exposure

to

computational

thinking

within

traditional

GIS

curricula.

Son

et

al.

(2021)

emphasized that meaningful learning in quantitative

disciplines is strengthened when students repeatedly

apply concepts through practical exercises, feedback,

and

reection.

Similarly,

Tucker

et

al.

(2023)

demonstrated

that

teaching

with

R

promotes

deeper

analytical

understanding

and

long-term

skill development when programming activities are

integrated progressively into coursework.

In

the

context,

students

were

already

familiar

with R for statistical purposes, yet they had limited

experience applying it to geospatial problems. This

distinction is important because spatial programming

involves

additional

competencies,

including

the

manipulation

of

spatial

objects,

coordinate

reference systems, and raster operations. Therefore,

reinforcing

R

training

through

complementary

workshops,

interactive

tutorials,

and

project-based

GIS exercises may improve both technical prociency

and

conceptual

understanding.

Such

strategies

could

help

students

overcome

initial

barriers

to

coding

while

simultaneously

strengthening

their

capacity

to

conduct

ecient,

reproducible,

and

scientically

rigorous

spatial

analyses,

which

are

increasingly required in agricultural, environmental,

and climate-related research (Pebesma and Bivand,

2023).

A major limitation of R and other programming

languages in GIS education is the steep learning curve

faced

by

university

students,

particularly

those

in

agronomical

engineering

programs.

Many

students

have

limited

prior

exposure

to

coding,

algorithmic

thinking,

and

data

structures,

which

can

generate

anxiety, frustration, and reduced motivation during

spatial

analysis

exercises

(Pavlenko

et

al.,

2022).

In

addition,

agronomical

students

often

prefer

eld-based

and

applied

activities,

making

abstract

programming

concepts

more

dicult

to

assimilate.

Challenges related to syntax, package management,

debugging,

and

coordinate

reference

systems

may

further

hinder

learning,

emphasizing

the

need

for

gradual,

practice-oriented,

and

interdisciplinary

teaching

strategies

(Tucker

et

al.,

2023;

Son

et

al.,

80

│

Volumen 19, Número 52, Septiembre-Diciembre 2026, pp. 72 - 83

2021).

Additional

Limitations

and

Future

Research

Beyond

the

instructional

confound

discussed

above,

several

other

methodological

limitations

must

be

acknowledged

when

interpreting

the

ndings. First, the sample size was relatively small

(n

=

52)

and

restricted

to

agronomic

engineering

students

at

a

single

institution,

which

may

limit

the generalizability of the results to other academic

programs,

skill

levels,

or

institutional

contexts.

Second, the evaluation was conned to a single raster

temperature extraction exercise; thus, performance

and software acceptance may dier when executing

other geospatial workows, such as vector processing,

spatial

interpolation,

or

complex

multi-criteria

analyses.

Third,

the

exercise

execution

order

was

not

randomized

or

counterbalanced;

participants

independently

decided

whether

to

start

with

R

or

QGIS,

introducing

potential

sequence,

fatigue,

or

cross-platform learning transfer eects. Finally, key

outcome variables specically process understanding

and

operational

diculty

were

measured

strictly

through

self-reported

survey

responses

rather

than

objective

pre-

and

post-instructional

knowledge

assessments.

While

self-perceived

clarity

provides

valuable insights into user condence and workow

transparency,

it

does

not

directly

capture

veried

conceptual

retention

or

skill

acquisition,

as

lower

operational

friction

in

scripted

environments

may

articially elevate perceived comprehension.

Consequently, future research should implement

randomized,

counterbalanced

designs

across

multi-institutional

cohorts,

employ

standardized

instructional materials, and incorporate objective pre-

and

post-exercise

conceptual

evaluations

to

isolate

intrinsic

software

capabilities

from

pedagogical

formatting

and

rigorously

quantify

actual

learning

gains

in

geospatial

education.

Longitudinal

studies

would also be valuable to assess long-term retention

and skill transfer.

Practical Recommendations for Educators

Based

on

our

ndings

and

their

limitations,

we

recommend that GIS educators adopt R Markdown

or

equivalent

scripted

tools

to

create

transparent,

reproducible,

and

modiable

workows,

while

providing

scaolded

scripts

that

students

can

progressively

modify

as

their

skills

develop.

A

blended

instructional

approach

combining

GUI

based QGIS for intuitive visualization with code based

R

for

automation

and

reproducibility

is

advocated

to

leverage

the

strengths

of

both

environments.

Supplementary coding support through workshops,

tutorials,

and

oce

hours

should

address

syntax

and debugging challenges, which students identied

as

primary

barriers.

Finally,

we

encourage

using

R

for

reproducibility

demanding

assignments,

such

as

research

projects

or

theses,

to

reinforce

best

practices in scientic computing and better prepare

students for data driven careers in agricultural and

environmental sciences.

V. CONCLUSION

This

study

demonstrates

that

the

R

Markdown

workow

was

associated

with

higher

completion

rates,

shorter

execution

times,

and

greater

perceived

understanding

compared

with

the

QGIS

handbook-based

approach.

However,

these

advantages

are

substantially

confounded

by

instructional

asymmetry

an

executable

script

versus

a

static

manual

precluding

any

claim

of

inherent software superiority. Instead, our ndings

underscore

the

pedagogical

value

of

scripted,

reproducible workows in GIS education. While 96%

of students would adopt R as a complementary tool,

primary preferences remained balanced (52% QGIS

vs. 48% R), supporting a blended instructional model

that combines intuitive visualization with code-based

automation.

From an educational perspective, we recommend

scaolded

R

Markdown

scripts

to

enhance

transparency and reproducibility, complemented by

workshops to address syntax and coding challenges

identied as key barriers. Critically, the main obstacle

was

insucient

prior

programming

training,

not

R's

geospatial

functionality

itself.

Future

studies

should

employ

counterbalanced,

multi-institutional

designs

with

standardized

materials

and

objective

assessments

to

disentangle

software

eects

from

pedagogical

formatting,

thereby

rigorously

quantifying actual learning gains in agricultural and

environmental spatial analysis education.

Acknowledgments

The

authors

are

grateful

to

the

students

of

│

81

Macancela

.

R or QGIS? A Case Study on Raster Analysis Eciency

agronomical engineering of the University of Cuenca

who

participated

in

completing

the

exercise

and

questionnaire.

Conict of interests

There is no type of conict of interest related to

the subject of the work.

Ethical consideration

In

accordance

with

the

ethical

principles

set

forth

in

the

Declaration

of

Helsinki,

electronic

informed consent was secured from all participants

before

administering

the

practical

exercise

and

questionnaire.

Students

received

comprehensive

information

regarding

the

study

objectives,

the

voluntary

character

of

their

participation,

and

the

complete

anonymization

of

their

responses.

Crucially, participants were notied that opting in or

out of the study would not inuence their academic

performance or standing in any manner.

Availability of deposited data:

The

R

markdown

and

data

questionaries

are

available

from

the

corresponding

author

upon

reasonable request.

VI. REFERENCES

Cuevas, E. H., Zaldivar, D., and Perez, M. (2025).

Impact

of

Programming

Languages

on

Learning

Performance. International Journal of Information and

Communication

Technology

Education,

21(1),

1–17.

https://doi.org/10.4018/IJICTE.371419

Giofandi,

E.,

Munibah,

K.,

Kraugusteeliana,

K.,

Novalinda,

A.,

and

Sekarrini,

C.

(2023).

The

Comparison

of

Vector

and

Raster

Data

for

The

Calculation

of

Landscape

Environment

Using

a

Geographic Information System Approach. IT Journal

Research

and

Development,

7(2),

209–219.

https://

doi.org/10.25299/itjrd.2023.10878

Graser,

A.,

Sutton,

T.,

y

Bernasocchi,

M.

(2025).

The

QGIS

project:

Spatial

without

compromise.

Patterns,

6(7),

101265.

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

patter.2025.101265

Hijmans,

R.

J.

(2020).

raster:

Geographic

Data

Analysis

and

Modeling.

R

package

version

3.4-5.

https://cran.r-project.org/package=raster

Hijmans, R. J. (2025). terra: Spatial data analysis

(R

package

version

1.8-60).

CRAN.

https://cran.r-

project.org/web/packages/terra/index.html

Knutas, A., Hynninen, T., and Hujala, M. (2021).

To Get Good Student Ratings should you only Teach

Programming Courses? Investigation and Implications

of

Student

Evaluations

of

Teaching

in

a

Software

Engineering

Context.

Proceedings

-

International

Conference on Software Engineering, 253–260. https://

doi.org/10.1109/ICSE-SEET52601.2021.00035

Li, X., Yue, J., Wang, S., Luo, Y., Su, C., Zhou, J.,

Xu, D., and Lu, H. (2024). Development of Geographic

Information System Architecture Feature Analysis and

Evolution Trend Research. Sustainability (Switzerland),

16(1). https://doi.org/10.3390/su16010137

Martin Gomez, S., and Bartolome Muñoz de Luna,

A.

(2023).

Systemic

Review

through

Bibliometric

Analysis with RStudio of Skills Learning to Favor the

Employability

of

Its

Graduates.

Trends

in

Higher

Education,

2(1),

101–122.

https://doi.org/10.3390/

higheredu2010007

Mladenović,

M.,

Krpan,

D.,

and

Mladenovi,

S.

(2017).

Learning

programming

from

scratch.

Turkish

Online

Journal

of

Educational

Technology,

2017(November

Special

Issue

INTE),

419–427.

https://www.semanticscholar.org/paper/Learning-

programming-from-Scratch-Mladenović-Krpan/7fc1d

e582f2fa9230b2c505ba07770adf347d5ef

Nguyen Tien, H., Nguyen Thi, H., and KOIKE, K.

(2019).

High

versatility

and

potential

of

spatial

data

analysis with R programming. Geoinformatics, 30(1),

3–14. https://doi.org/10.6010/geoinformatics.30.1_3

Parra,

M.

I.,

Sanjuán,

E.

L.,

Robustillo,

M.

C.,

and Pizarro, M. M. (2023). Using R for teaching and

research. http://arxiv.org/abs/2306.12200

Paucar-Curasma,

R.,

Silveira,

I.

F.,

Rondon,

D.,

Robles,

Z.

M.

L.,

and

Porras-Ccancce,

L.

E.

(2023).

Analysis of the Teaching of Programming and Evaluation

of

Computational

Thinking

in

Recently

Admitted

Students at a Public University in the Andean Region

of Peru. CEUR Workshop Proceedings, 3353, 101–110.

82

│

Volumen 19, Número 52, Septiembre-Diciembre 2026, pp. 72 - 83

https://dspace.mackenzie.br/items/5b561733-5c90-

44c6-8385-49bd7ace46b9

Pavlenko,

L.,

Pavlenko,

M.,

Khomenko,

V.,

and

Mezhuyev, V. (2022). Application of R Programming

Language

in

Learning

Statistics.

January,

62–72.

https://doi.org/10.5220/0010928500003364

Pebesma,

E.

(2018).

Simple

features

for

R:

Standardized

support

for

spatial

vector

data.

The

R

Journal, 10(1), 439–446.

Pebesma,

E.,

and

Bivand,

R.

(2023).

Spatial

data

science:

With

applications

in

R.

Chapman

and

Hall/

CRC.

Portella-Cleves,

J.

E.,

and

Rodríguez-Hernández,

A.

(2024).

Enhancing

Programming

Education

with

an

Active

Learning

Plan

and

Articial

Intelligence

Integration.

Revista

Facultad

de

Ingeniería (Rev. Fac. Ing.), 33(67), 2024. https://doi.

org/10.19053/01211129.v33.n67.2024.16328

Poynton, C. (2003). 1 - Raster images. In C. B. T.-

D.

V.

and

H.

Poynton

(Ed.),

The

Morgan

Kaufmann

Series

in

Computer

Graphics

(pp.

1–16).

Morgan

Kaufmann.

https://doi.org/https://doi.org/10.1016/

B978-155860792-7/50067-7

R

Core

Team.

(2026).

R:

A

language

and

environment

for

statistical

computing.

R

Foundation

for Statistical Computing. https://www.r-project.org/

Rasul, A., Ibrahim, S., Onojeghuo, A. R., y Balzter,

H.

(2020).

A

trend

analysis

of

leaf

area

index

and

land surface temperature and their relationship from

global

to

local

scale.

Land,

9(10),

1–17.

https://doi.

org/10.3390/land9100388

Saqr,

M.,

and

López-Pernas,

S.

(2024).

Learning

Analytics

Methods

and

Tutorials:

A

Practical

Guide

Using

R.

Learning

Analytics

Methods

and

Tutorials:

A

Practical

Guide

Using

R,

June,

1–736.

https://doi.

org/10.1007/978-3-031-54464-4

Scholten, H. J. (1990). Application of geographical

information systems in the Netherlands. Kartograsch

Tijdschrift,

16(4),

27–36.

https://pubmed.ncbi.nlm.

nih.gov/1949884/

Singla,

S.,

Eldawy,

A.,

Diao,

T.,

Mukhopadhyay,

A.,

and

Scudiero,

E.

(2021).

Experimental

study

of

big

raster

and

vector

database

systems.

Proceedings

-

International

Conference

on

Data

Engineering,

2021-April,

2243–2248.

https://doi.org/10.1109/

ICDE51399.2021.00231

Son, J. Y., Blake, A. B., Fries, L., and Stigler, J. W.

(2021).

Modeling

First:

Applying

Learning

Science

to the Teaching of Introductory Statistics. Journal of

Statistics

and

Data

Science

Education,

29(1),

4–21.

https://doi.org/10.1080/10691898.2020.1844106

Steiniger, S., De La Fuente, H., Fuentes, C., Barton,

J.,

and

Muñoz,

J.

C.

(2017).

Building

a

geographic

data repository for urban research with free software

- Learning from Observatorio.Cedeus.Cl. International