Ind. J . o Mod. Res. and Re . PEER-REVIEWED JOURNAL Volume 3 Issue 11 [No ] 2025
41
© 2025 Nomsa C. C. Kamgwi a, Shalu Gup a. This is an open-access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0
In e na ional License (CC BY 4.0).h ps://c ea i ecommons.o g/licenses/by/4.0/
Indian Jou nal o
Mode n Resea ch and Re iews
This Jou nal is a membe o he ‘Commi ee on Publica ion E hics’
Online ISSN:2584-184X
Resea ch Pape
A Sys ema ic Re iew o Machine Lea ning Algo i hms o
Classi ica ion: Gene al App oaches and En i onmen al Applica ions
Nomsa C. C. Kamgwi a 1, Shalu Gup a 2*
1 S uden , Depa men o Compu e Applica ions, Gu u Kashi Uni e si y, Talwandi Sabo, Punjab, India
2 Associa e P o esso , Depa men o Compu e Applica ions, Gu u Kashi Uni e si y, Talwandi Sabo, Punjab, India
Co esponding Au ho : *Shalu Gup a DOI: h ps://doi.o g/10.5281/zenodo.17722101
2
ABSTRACT
Manusc ip In o.
The ield o Machine Lea ning has seen apid ad ancemen om 2022 o 2025 due o mo e
cu ing-edge compu a ional ools, hyb id models and imp o ed speci ied echniques. This
e iew has been w i en o assess he widely used classi ica ion algo i hms, such as adi ional,
ensemble-based and deep lea ning. I e alua es hei pe o mance in p ac ical applica ions. The
sea ch co e ed i e open lea ning da abases, which a e: Google Schola , Seman ic Schola ,
a Xi , DOAJ and Resea chGa e. 57 s udies published be ween 2022 and 2025 me he selec ion
c i e ia. Findings show ha Random Fo es s and XGBoos a e e ec i e o s uc u ed da ase s,
and CNNs and ans o me s a e mo e sui able o uns uc u ed da ase s. Hyb id deep lea ning
ensembles a e mo e s able as hey can cap u e spa ial and empo al pa e ns. This e iew
p o ides a summa y o he esul s, including a compa ison able and an ou line o a eas ha
equi e u he wo k.
✓ ISSN No: 2584- 184X
✓ Recei ed: 08-9-2025
✓ Accep ed: 29-10-2025
✓ Published: 26-11-2025
✓ MRR:3(11):2025;41-47
✓ ©2025, All Righ s Rese ed.
✓ Pee Re iew P ocess: Yes
✓ Plagia ism Checked: Yes
How To Ci e his A icle
Kamgwi a NCC, Gup a S. A
sys ema ic e iew o machine lea ning
algo i hms o classi ica ion: gene al
app oaches and en i onmen al
applica ions. Indian J Mod Res Re .
2025;3(11):41-47.
KEYWORDS: Machine lea ning; Classi ica ion algo i hms; Ensemble lea ning; Deep lea ning; T ans o me s; En i onmen al
classi ica ion; Sys ema ic e iew; Open-access da abases.
1. INTRODUCTION
Machine lea ning (ML) classi ica ion algo i hms play an
impo an ole in da a-d i en decision-making ac oss ields such
as heal hca e, inance, cybe secu i y, ag icul u e, and
en i onmen al science. Be ween 2022 and 2025, esea ch on
classi ica ion accele a ed due o la ge da ase s, imp o ed
compu a ional ools, and expanded open-sou ce ML lib a ies.
Recen algo i hms, mainly ans o me -based models, ha e
enhanced he abili ies o adi ional classi ica ion app oaches.
E en hough he e a e a lo o e iew pape s ou he e, he ocus
is on a single ield o speci ic ela ed algo i hms. Some s udies
combine gene al algo i hm pe o mance wi h a deepe
unde s anding o eal-wo ld applica ions.
En i onmen al classi ica ion has become inc easingly ele an
wi h clima e change ( ain all p edic ion, pollu ion ca ego isa ion,
and wea he - ela ed isk assessmen ), ye esea ch in his a ea is
s ill sca e ed [1,2].
The objec i e o he e iew is o add ess hese gaps by:
1. P esen ing an upda ed (2022–2025) sys ema ic e iew o
commonly used ML classi ica ion algo i hms
Ind. J . o Mod. Res. and Re . PEER-REVIEWED JOURNAL Volume 3 Issue 11 [No ] 2025
42
© 2025 Nomsa C. C. Kamgwi a, Shalu Gup a. This is an open-access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0
In e na ional License (CC BY 4.0).h ps://c ea i ecommons.o g/licenses/by/4.0/
2. Compa ing pe o mance ends ac oss s uc u ed, image,
ex , and senso -based da ase s
3. P o iding en i onmen al classi ica ion as a p ac ical
example
4. Adop ing open esea ch p ac ices ha p omo e he sha ing
o da a, me hods, and esul s
5. P esen ing ables and summa ised indings o esea che s
and pos g adua e s uden s
ML classi ica ion algo i hms all in o h ee g oups:
1.1 T adi ional Classi ica ion Algo i hms
Common adi ional algo i hms a e Logis ic Reg ession and
Suppo Vec o Machines (SVM), Naï e Bayes, k-Nea es
Neighbou s (k-NN), and Decision T ees. They wo k well wi h
s uc u ed da a and small- o-medium da ase s. Cu en esea ch
explo es imp o emen s such as ke nelised SVMs, p o iding
imp o ed cla i y and unde s anding [6].
1.2 Ensemble-Based Algo i hms
Ensemble models a e mo e powe ul and a e c ea ed by pooling
he esul s om se e al indi idual p edic i e models. Common
models a e Random Fo es , AdaBoos , G adien Boos ing,
Ca Boos , and XGBoos . As o 2022, XGBoos and Ca Boos
gained ac ion o handling common issues e icien ly, like
handling missing da a, ca ego ical a iables, and class imbalance
[7].
1.3 Deep Lea ning and T ans o me -Based Algo i hms
Deep lea ning app oaches, speci ically CNNs, RNNs, LSTMs,
and ans o me s, a e b oadly applied o uns uc u ed da a,
which encompasses images, audio, and ex [8].
T ans o me s we e i s de eloped o Na u al Language
P ocessing (NLP). Howe e , hey a e now used in ime-se ies
and en i onmen al applica ions, showing s ong pe o mance in
ecognising complex pa e ns [1].
1.4 En i onmen al Classi ica ion as a Case Example
En i onmen al da ase s ypically include:
• Nonlinea pa e ns
• High empo al a iabili y
• Spa ial dependencies
• Noisy o missing obse a ions
These ea u es make en i onmen al classi ica ion a sui able es
case o algo i hm obus ness. Recen s udies demons a e ha
hyb id CNN–LSTM models, XGBoos -based ea u e selec ion
pipelines, and ans o me a chi ec u es pe o m well in
classi ying ain all in ensi y, ai quali y, and d ough se e i y
[10].
1.5 Pu pose o This Re iew
This s udy analy ically e iews machine lea ning classi ica ion
algo i hms using open-access li e a u e a ailable om 2022 o
2025, e alua ing s anda d, widely-used machine lea ning
me hods in classi ying en i onmen al da a. The e iew aims o
cla i y:
• Changing dynamics in classi ica ion ends
• E ec i e hyb ids wi hin he ield
• Cu en limi a ions and a eas o u u e s udy
2. METHODOLOGY
This e iew ollowed a p o ocol modelled on he PRISMA 2020
guidelines, which is a widely ecognised amewo k designed o
imp o e anspa ency o sys ema ic e iews. The me hodology
makes su e ha he esea ch p ocess is clea and can be
ep oduced. This allows o he esea che s o eplica e he same
s eps and con i m he indings.
The Wea he AUS da ase is used o p edic wea he e en s in
Aus alia as i p esen s a signi ican class imbalance. The e a e
mo e ins ances o ‘No Rain’ han ‘Rain’. This imbalance aises
di icul ies o machine lea ning models, which end o p e e
he majo i y class. This esul s in exagge a ed accu acy sco es
while unde pe o ming in classi ying ac ual ain e en s. As a
esul , he F1 sco e, which balances p ecision and ecall, educes
o he mino i y class [4].
This p oblem is well-known in en i onmen al p edic ion
esea ch, hough a e bu c i ical e en s like ain all a e
occasional and highly a iable, making hem ha de o model
e ec i ely. To esol e imbalance e ec s, s anda d p ep ocessing
echniques such as s a i ied spli ing, no malisa ion we e
applied. Howe e , he imbalance emains a key ac o
in luencing he expe imen al ou comes and mus be aken in o
conside a ion when in e p e ing classi ie pe o mance [5].
2.1 Sea ch S a egy
The li e a u e sea ch was pe o med exclusi ely on open-access
pla o ms. The ollowing da abases we e used:
• Google Schola : p ima y sea ch engine
• Seman ic Schola : links o open-access e sions
• a Xi : p ep in s in machine lea ning and en i onmen al
modelling
• Di ec o y o Open Access Jou nals (DOAJ): pee - e iewed
open-access jou nals
• Resea chGa e: p ep in s o accep ed manusc ip s uploaded
by au ho s
Sea ch keywo d
Keywo ds and Boolean ope a o s included:
• “Machine lea ning classi ica ion”
• “Supe ised lea ning models”
• “Classi ica ion algo i hms e iew”
• “Deep lea ning classi ie ”
• “T ans o me classi ica ion model”
• “En i onmen al classi ica ion” OR “wea he classi ica ion”
• “Rain all classi ica ion” AND “machine lea ning”
Sea ch ime ame
Only s udies published be ween Janua y 2022 and Janua y 2025
we e conside ed.
2.2 Inclusion C i e ia
S udies we e included i hey me all o he ollowing:
• Published 2022–2025
• F ee PDFs and open-access p ep in
• Applied and discussed machine lea ning classi ica ion
algo i hms
Ind. J . o Mod. Res. and Re . PEER-REVIEWED JOURNAL Volume 3 Issue 11 [No ] 2025
43
© 2025 Nomsa C. C. Kamgwi a, Shalu Gup a. This is an open-access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0
In e na ional License (CC BY 4.0).h ps://c ea i ecommons.o g/licenses/by/4.0/
• Repo ed pe o mance me ics such as accu acy, F1-sco e,
p ecision, ecall, o AUC
• W i en in English
• Pee - e iewed OR p ep in om a Xi wi h es ablished
au ho c edibili y
• Fo a domain example: s udies applying ML o
en i onmen al classi ica ion asks
2.3 Exclusion C i e ia
S udies we e excluded i hey:
• Requi ed paid access (e.g., publishe paywalls)
• Focused solely on eg ession, clus e ing, o ein o cemen
lea ning
• Did no include any classi ica ion model
• Lacked me hodological de ail
• P o ided no quan i a i e esul s
• Duplica e en ies
• Non-English o inaccessible manusc ip s
2.4 S udy Selec ion P ocess (PRISMA Desc ip ion)
A o al o 265 eco ds we e ound ac oss all ee-access
da abases. A e emo ing duplica es, checking i les and
abs ac s, 121 s udies we e le . A de ailed assessmen emo ed
a icles lacking ele ance (e.g., missing me ics, unclea
me hodology). This le [8]:
• Picked s udies: 57 (2022–2025)
PRISMA-S yle Flow Desc ip ion
Iden i ica ion
• Reco ds ound h ough open-access da abases: 265
• Duplica es emo ed: 56
Sc eening
• Reco ds sc eened ia i le/abs ac : 209
• Reco ds excluded o i ele ance: 88
Eligibili y
• De ailed a icles assessed: 121
• De ailed excluded (inaccessible, no classi ica ion, lacked
me ics): 64
Included
• S udies included in he inal e iew: 57
2.5 Da a Ex ac ion
Fo each included pape , he ollowing in o ma ion was
e iewed:
• Publica ion yea
• Algo i hm ypes used
• Da ase cha ac e is ics
• Field o s udy (gene al s. en i onmen al)
• E alua ion me ics
• Model compa ison esul s
• Key con ibu ions o inno a ions
• Limi a ions
Ex ac ion was done manually o a oid bias and ensu e
accu acy.
2.6 Quali y Assessmen
The quali y o included s udies was e alua ed acco ding o
se e al key c i e ia:
• Cla i y o da a desc ip ion
• A ailabili y o he da ase
• Explana ion o p ep ocessing s eps
• Valida ion echnique
• Repo ing o mul iple me ics
S udies sco ing below 50% on quali y indica o s we e
excluded.
3. RESULTS
This sec ion shows he indings om he 57 included s udies and
combines an expe imen al sec ion conduc ed by he esea che o
add o he li e a u e-based indings. This expe imen is o check
commonly used classi ica ion algo i hms on a public da ase o
measu e hei pe o mance.
3.1 Li e a u e-Based Resul s: Gene al Classi ica ion
Algo i hms
Analysis o he s udies used (2022–2025) shows clea ends:
• Ensemble models (XGBoos , Ca Boos , Random Fo es )
always ou pe o m adi ional algo i hms on s uc u ed
da ase s.
• CNNs a e dominan o image- and senso -based
classi ica ion asks.
• T ans o me s gi e he bes esul s on ex , sa elli e image y,
and ime-se ies en i onmen al classi ica ion.
Hyb id models combining deep lea ning wi h ensemble
app oaches eme ged as he highes -pe o ming ca ego y in
complex en i onmen al asks.
Table 1: Summa y o Gene al Classi ica ion Algo i hm Pe o mance (2022–2025)
Algo i hm
Typical Accu acy Range
Bes Da a Type
Key S eng h
Common Limi a ion
Example Open-Access S udy
Logis ic Reg ession
70–85%
Tabula
In e p e able
Limi ed o linea ends
Khan e . al. (2022) [10]
IGWO SVM
80–98.75%
Tabula /Image
Good o small da ase s
High aining cos
Shen e . al. (2023) [11]
Random Fo es
85–96%
Tabula
Robus
Memo y-hea y
Ahmed e al. [12]
XGBoos
88–98%
Tabula
Handles missing da a
Requi es uning
Wu & Zhang (2023) [13]
Ca Boos
87–97%
Ca ego ical
No encoding needed
Slowe
Li e al. [14]
CNN
90–99%
Image/Senso
Bes o spa ial da a
Needs la ge da a
Singh e al. [15]
LSTM/GRU
83–95%
Time Se ies
Models’ long sequences
Slow aining
Ogundele e al. [16]
T ans o me
92–99%
Tex /Image/TS
S a e-o - he-a
High cos
Chen e al. [17]
Ind. J . o Mod. Res. and Re . PEER-REVIEWED JOURNAL Volume 3 Issue 11 [No ] 2025
44
© 2025 Nomsa C. C. Kamgwi a, Shalu Gup a. This is an open-access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0
In e na ional License (CC BY 4.0).h ps://c ea i ecommons.o g/licenses/by/4.0/
3.2 Li e a u e-Based Resul s: En i onmen al Classi ica ion
En i onmen al applica ions included ain all p edic ion, lood
isk classi ica ion, d ough se e i y analysis, and ai quali y
ca ego isa ion.
Key de elopmen s:
• CNN–LSTM hyb ids excel in asks ha ha e spa ial and
empo al dependencies.
• XGBoos pe o ms bes o ai -quali y senso da a due o he
abula s uc u e.
• T ans o me s (especially ime-se ies a ian s) a e eme ging
leade s in ain all and pollu ion classi ica ion.
Table 2: En i onmen al Classi ica ion Algo i hms and Findings (2022–2025)
S udy
Task
Bes Pe o ming Model
Me ic (Repo ed)
Key Insigh
Wang e al., 2022
[18]
Rain all e en classi ica ion
CNN–LSTM
93.6% accu acy
Hyb id deep models cap u e spa ial + empo al ain all
pa e ns e ec i ely.
Cheng e al., 2023
[19]
Ai quali y le el classi ica ion
XGBoos
95–97% accu acy
G adien boos ing handles senso noise and missing alues
well
Huang e al., 2024
[20]
Flood suscep ibili y
classi ica ion
T ans o me -based
encode
AUC = 0.92
T ans o me a en ion imp o es long- e m hyd ological
dependency modelling
Rahman e al.,
2023 [21]
Land co e & sa elli e image
classi ica ion
CNN
98% accu acy
CNNs bes ex ac spa ial spec al pa e ns in emo e
sensing images
Li e al., 2025 [22]
D ough se e i y
classi ica ion
Random Fo es
79.9% accu acy
RF emains obus in noisy clima e a iables; s ong
ea u e-impo ance in e p e a ion.
3.3 Expe imen al Componen (Resea che -Execu ed Tes )
To suppo indings om he li e a u e, a small expe imen al
e alua ion was conduc ed using he Wea he AUS da ase , an
openly accessible me eo ological da ase con aining daily
wea he obse a ions om mul iple Aus alian egions.
The da ase has en i onmen al a iables such as empe a u e,
humidi y, ain all, a mosphe ic p essu e, cloud co e ,
e apo a ion, wind cha ac e is ics, and he a ge a iable
RainTomo ow, which is a bina y classi ica ion label. This
makes i sui able o p o ing he accu acy o he ends obse ed
in en i onmen al classi ica ion esea ch.
The ollowing classi ica ion algo i hms we e selec ed o
compa ison, e lec ing hose mos common in he 2022–2025
li e a u e:
• Logis ic Reg ession
• Suppo Vec o Machine (SVM)
• Random Fo es Classi ie
• XGBoos Classi ie
This expe imen is no in ended o p opose new models; a he ,
i s pu pose is o:
• Demons a e anspa ency and ep oducibili y
• P o ide an empi ical benchma k aligned wi h he sys ema ic
e iew
• Valida e whe he he Wea he AUS da ase exhibi s he same
ends epo ed in Tables 1 and 2
All p ep ocessing s eps, handling missing alues, encoding
ca ego ical a iables, spli ing in o aining/ es ing se s, and
applying ea u e scaling, ollowed s anda d p ocedu es used in
published en i onmen al classi ica ion s udies.
3.3.1 Expe imen al Resul s
Table 3: Expe imen al Pe o mance o Classi ica ion Algo i hms on he Wea he AUS Da ase
Model
Accu acy
F1 Sco e
Logis ic Reg ession
0.839513
0.550685
SVM (RBF Ke nel)
0.847212
0.548930
Random Fo es
0.848694
0.585839
XGBoos
0.849789
0.586210
Ind. J . o Mod. Res. and Re . PEER-REVIEWED JOURNAL Volume 3 Issue 11 [No ] 2025
45
© 2025 Nomsa C. C. Kamgwi a, Shalu Gup a. This is an open-access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0
In e na ional License (CC BY 4.0).h ps://c ea i ecommons.o g/licenses/by/4.0/
Figu e 1: Accu acy compa ison o he ou classi ica ion models on he Wea he AUS da ase
Figu e 2: F1-sco e compa ison showing he pe o mance di e ences among he models on he Wea he AUS da ase
Ind. J . o Mod. Res. and Re . PEER-REVIEWED JOURNAL Volume 3 Issue 11 [No ] 2025
46
© 2025 Nomsa C. C. Kamgwi a, Shalu Gup a. This is an open-access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0
In e na ional License (CC BY 4.0).h ps://c ea i ecommons.o g/licenses/by/4.0/
3.4 In e p e a ion o Expe imen al Findings
The Wea he AUS da ase was used o e alua e he pe o mance
o common classi ica ion algo i hms o ain all p edic ion.
Accu acy alues o all models anged na owly (0.839–0.850),
e lec ing he inhe en di icul y o p edic ing ain all om
me eo ological a iables.
He e a e he esul s o hei pe o mance:
• XGBoos achie ed he highes accu acy (0.8498) and F1
sco e (0.5862)
• Random Fo es (accu acy 0.8487; F1 0.5858)
This is consis en wi h s udies showing ensemble me hods
e ec i ely model nonlinea in e ac ions and educe o e i ing in
abula me eo ological da a. Logis ic Reg ession sco ed lowes
(accu acy 0.8395; F1 0.5507), while SVM was sligh ly be e
(accu acy 0.8472; F1 0.5489), wi h F1 sco es a ec ed by he
da ase ’s class imbalance.
All models sco ed modes F1 sco es (0.55–0.59), which is
expec ed due o he p edominance o “No Rain” ins ances. The
imbalance in he da a educes he model’s abili y o co ec ly
iden i y ain all, as i ends o p edic he mo e common
ou come.
Ensemble me hods con inue o be he mos e ec i e o abula
en i onmen al classi ica ion, bu gains o e adi ional models
a e inc emen al. The esul s also shed ligh on da a challenges
such as imbalance, noise and nonlinea a mosphe ic beha iou
ha cons ain classi ie pe o mance.
4. DISCUSSION AND CONCLUSION
This sys ema ic e iew and expe imen al e alua ion es s how
machine lea ning classi ica ion algo i hms pe o m on gene al
and en i onmen al da ase s. Re iewing he 57 s udies om
2022–2025 e ealed clea pa e ns ha show ha :
• Ensemble models pe o m well on abula da a
• Deep lea ning is bes on image and spa io empo al da a
• T ans o me -based a chi ec u es a e inc easingly used o
complex en i onmen al p oblems.
This shows a shi owa d hyb id and ensemble-deep app oaches
o eal-wo ld en i onmen al applica ions.
All he es ed models had kind o he same accu acy on he
Wea he AUS da ase ; ensemble me hods (Random Fo es and
XGBoos ) pe o med bes , especially in he F1 sco e. This
ma ches he exis ing da a, which shows ensembles handle noisy,
nonlinea and a ied me eo ological da a e ec i ely.
The expe imen s also highligh ed he limi s o s anda d machine
lea ning o ain all. The mode a e F1 sco es (0.55–0.59) show
how di icul i is o p edic ainy days, which a e less common
in he da ase . Techniques such as SMOTE, cos -sensi i e
lea ning, o empo al models can be able o help imp o e
p edic ion accu acy.
Classical machine lea ning can p o ide easonable wea he
p edic ions, bu mo e accu a e esul s need ad anced models,
iche ea u es, o longe - e m da ase s. Fu u e wo k could look
in o ans o me -based ime se ies models, hyb id CNN–LSTM
a chi ec u es o imbalance-awa e aining. Expe imen ing wi h
deep lea ning app oaches in-dep h could e eal addi ional
s eng hs and limi a ions o cu en me hods.
REFERENCES
1. Wang R, Ma L, He G, Johnson BA, Yan Z, Chang M,
Liang Y. T ans o me s o emo e sensing: A sys ema ic
e iew and analysis. Senso s. 2024;24(11):3495.
2. Amanambu AC, Adika i S. Hyd ological d ough
o ecas ing using a deep ans o me model. Wa e .
2022;14(22):3611.
3. Yang J, Tian Y, Wu CH. Ai quali y p edic ion and anking
assessmen based on Boo s ap-XGBoos algo i hm and
o dinal classi ica ion models. A mosphe e. 2024;15(8):925.
4. Sa asa-Cabezuelo A. P edic ion o ain all in Aus alia
using machine lea ning. In o ma ion. 2022;13(4):163.
5. Majid A, Saga BSD. Rain all p edic ion using machine
lea ning algo i hms: A e iew. Ea h Science In o ma ics.
2023;16:2341-2360.
6. Ko sian is SB. Supe ised machine lea ning: A e iew o
classi ica ion echniques. In o ma ics. 2023;10(1):12.
7. Hossain M, Rahman MM. An imp o ed ke nel-based
suppo ec o machine o classi ica ion asks. Appl Sci.
2022;12(14):6841.
8. Khan A, Sohail A, Zahoo a U, Qu eshi A. A su ey o he
ecen a chi ec u es o deep con olu ional neu al ne wo ks.
A i In ell Re . 2022;55:469–548.
9. Islam F, Khan MZ, Akh e U, Aslam S. Imbalanced da a
p ep ocessing echniques o machine lea ning: A
sys ema ic mapping s udy. PLOS ONE. 2023.
10. Khan M, Yeo K, Hossen J. Wea he p edic ion using
logis ic eg ession. A mosphe e. 2022;13(4):596.
11. Shen J, e al. Enhanced SVM classi ica ion using imp o ed
g ey wol op imisa ion. Appl Sci. 2023.
12. Ahmed F, e al. Random o es -based ai quali y p edic ion.
Sus ainabili y. 2023.
13. Wu L, Zhang H. XGBoos o ain all p edic ion in sma
ag icul u e. IEEE Access. 2023.
14. Li X, e al. Ca Boos -based d ough se e i y classi ica ion.
Wa e . 2023.
15. Singh P, e al. CNN-based lood de ec ion om sa elli e
images. Remo e Sens. 2022.
16. Ogundele L, e al. LSTM-based ain all o ecas ing.
En i on Da a Sci. 2023.
17. Chen Y, e al. Tempo al ans o me o ai quali y
o ecas ing. a Xi ; 2024.
18. Wang H, Li X, Zhang Q, Chen Y. Rain all e en
classi ica ion using a hyb id CNN–LSTM deep lea ning
model. Wa e . 2022;14(19):3022.
19. Cheng F, Zhao Y, Liu H, Sun J. Ai quali y classi ica ion
based on XGBoos and in eg a ed ea u e enginee ing.
A mosphe e. 2023;14(5):800.
Ind. J . o Mod. Res. and Re . PEER-REVIEWED JOURNAL Volume 3 Issue 11 [No ] 2025
47
© 2025 Nomsa C. C. Kamgwi a, Shalu Gup a. This is an open-access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0
In e na ional License (CC BY 4.0).h ps://c ea i ecommons.o g/licenses/by/4.0/
20. Huang Y, Wang S, Lin C, Du P. T ans o me -based lood
suscep ibili y mapping using mul i-sou ce en i onmen al
da a. Remo e Sens. 2024;16(2):351.
21. Rahman MM, Hassan QK, Dewan A. Deep CNN models
o mul iclass land co e classi ica ion om sa elli e
image y. PLOS ONE. 2023;18(3):e0282643.
22. Li M, Yao Y, Feng Z, Ou M. Hyd ological d ough
p edic ion and i s in luencing ea u es analysis based on a
machine lea ning model. Na Haza ds Ea h Sys Sci. 2025;
25:4299–316.
C ea i e Commons (CC) License
This a icle is an open-access a icle dis ibu ed unde he e ms and
condi ions o he C ea i e Commons A ibu ion (CC BY 4.0) license. This
license pe mi s un es ic ed use, dis ibu ion, and ep oduc ion in any
medium, p o ided he o iginal au ho and sou ce a e c edi ed.
