Analyzing Behavioral and Population Dynamics of House Sparrows Using ARIMA, Classification, and Clus

International Research Journal of Engineering and Technology (IRJET) e-ISSN:2395-0056

Volume: 12 Issue: 04 | Apr 2025 www.irjet.net p-ISSN:2395-0072

Analyzing Behavioral and Population Dynamics of House Sparrows Using ARIMA, Classification, and Clustering Models

Omkar Singh1, Amit Kumar Pandey2, Giridhar Balakrishnan3, Jagmohan Bal 4 ,

1HOD of MSc Data Science, 2 Assistant Professor, 3,4 PG Student (MSc Data Science), Thakur College of Science and Commerce

Thakur Village, Kandivali (East), Mumbai-400101, Maharashtra, India ***

Abstract:

Once a common sight in both urban and rural settings, house sparrows (Passer domesticus) have seen substantial populationchangesovertime.Conservationandecologicalbalancedependonourabilitytocomprehendtheirpopulation dynamics and behavioral adaptations. This work analyzes house sparrow populations across various geographic and temporal variations using machine learning approaches, such as clustering to identify important environmental parameters,classificationforhabitat-basedpopulationanalysis,andARIMAfortime-seriesforecasting.

Time-dependent population changes are evaluated by analyzing the dataset covering 2015–2025. While classification modelslikeRandomForestareusedtoclassifysparrowpopulationsaccordingtoenvironmentalcircumstances,ARIMAis used to predict future population trends. Different population groupings and environmental factors affecting population dynamicscanbeidentifiedwiththeuseofclusteringtechniquessuchasK-MeansandHierarchicalClustering.

Thefindingsshowthatclusteringidentifiesimportantpatternsinsparrowhabitats,classificationmodelscorrectlyidentify populationtrendsbasedonenvironmentalparameters,andARIMApredictspopulationchanges.Byhelpingpolicymakers create data-driven plans for preserving bird biodiversity, this research advances our understanding of house sparrow conservationefforts.

Index Terms:

HouseSparrowTrends,EnvironmentalFactors,PredictiveModeling,TimeSeriesForecasting,AvianPopulationDynamics, HabitatClustering,GeospatialAnalysis,ARIMAForecasting,SpeciesObservationData,K-MeansClustering,RandomForest Classification,EcologicalDataAnalysis,eBirdDatasetProcessing,SatelliteImage-BasedAnalysis.

Introduction:

Onceaspeciesthatthrivedinbothurbanandruralsettings,housesparrows(Passer domesticus)haveexperiencedsharp population declines in recent years. Urbanization, climate change, and habitat loss are among the primary causes of this reduction. While traditional ecological studies rely on field surveys and observational data, these approaches often lack scalability and predictive accuracy. Machine learning offers a data-driven approach to analyzing large-scale avian population data; however, research in this area remains limited, with most studies focusing on direct field observations ratherthanpredictivemodelling.

Several key challenges exist in house sparrow population analysis. There is a limited application of artificial intelligence andmachinelearningmodelsforstudyinglong-termpopulationdynamics.Additionally,thelackofintegrationoftemporal trends makes it difficult to accurately forecast future sparrow populations. Data-driven methods for classifying habitatbased population fluctuations are underutilized, and there is a need for a comprehensive examination of environmental factorsinfluencingsparrowdistribution.

To address these gaps, this study applies machine learning techniques, including Random Forest for habitat-based classification,ARIMAfortime-seriesforecasting,andK-Meansclusteringtoidentifyenvironmentalinfluencesonsparrow populations.Thedataset,coveringtheyears2015–2025,enablesanin-depthanalysisofpopulationtrendsacrossdifferent geographicalandtemporalscales.

The primary objectives of this research are: developing a predictive model using ARIMA to anticipate future sparrow population trends; classifying populations based on habitat conditions using machine learning algorithms, and utilizing

International Research Journal of Engineering and Technology (IRJET) e-ISSN:2395-0056

Volume: 12 Issue: 04 | Apr 2025 www.irjet.net p-ISSN:2395-0072

clustering techniques to identify key environmental factors influencing population dynamics. By integrating these approaches,thisstudysupportsdata-drivenconservationstrategies,providingecologistsandpolicymakerswithvaluable insightsforpreservinghousesparrowpopulationsandunderstandingtheirecologicaladaptations.

Literature Review:

Numerous studies have looked at the causes of the house sparrow (Passer domesticus) population decline, which has ecologists more concerned. Numerous contributory variables have been discovered by researchers, including increased pesticideuse,habitatdestruction,communicationtowerelectromagneticradiation,andfasturbanization.(1)Sharmaetal. highlight how sparrow populations have been gradually declining as a result of modern construction designs and infrastructurethatdrasticallylimitbreedingoptions.Likewise,Shansazetal.(2)drawattentiontothedetrimentaleffects that chemical exposure and environmental pollution have had on sparrow populations in both urban and rural settings.modelingtocomprehendsparrowpopulationdynamics.

Accordingtorecentdevelopmentsinecologicaldataanalysis,combiningclassificationmodels(RandomForest),clustering (K-Means),andtime-seriesforecasting(ARIMA)canprovidegreaterinsightsintothedynamicsofspeciespopulations.In order to categorize habitat types and forecast future population changes, previous research has mostly depended on mapping geographic distributions without utilizing unsupervised learning approaches. By using geospatial analysis tools oneBirdobservationaldata,thisstudyseekstoclosethisgapandprovideamorethoroughanddata-drivenapproachto housesparrowconservation.

Methodology:

4.1

TheoreticalAnalysis

Three main analytical techniques are used in this work to look at house sparrow population trends. First, historical sparrow population data is analyzed and future trends are predicted by time series forecasting utilizing the ARIMA (AutoRegressive Integrated Moving Average) model. This facilitates the identification of possible decline phases and enablesagreaterknowledgeofpopulationvariationsthroughouttime.

Second, K-Means clustering is used for habitat clustering through geospatial analysis. This method groups sparrow habitats according to their geographic locations and environmental characteristics. This method facilitates the identificationofareaswithhighandlowpopulationdensities,allowingformorefocusedconservationefforts.

Finally, the Random Forest classification model is used to classify species based on observational data. This aids in the analysis of sparrow sightings and the identification of important environmental variables affecting population dispersal. Thisstudyoffersathoroughanddata-drivenecologicalanalysisthatgoesbeyondconventionalconservationassessments bycombiningthesemethodologies.

4.2

Software&ToolsUsed

This study uses a range of tools and libraries for modelling, data processing, and visualization. Pandas and NumPy are utilizedforfeaturetransformation,missingvaluemanagement,andeffectivedatasetstructuring.Visualrepresentationsof populationtrends,clusterdistribution,andforecastingoutcomesareproducedwiththehelpofMatplotlibandSeaborn.

For machine learning applications, Scikit-learn is applied for classification and clustering, while Statsmodels is used for ARIMA-based time series forecasting. The mapping of sparrow distribution patterns onto satellite data is made possible viaGeopandasandContextily,whichfacilitategeographicvisualization.

Visual Studio Code is used for the entire implementation, which guarantees modular experimentation, structured model training,andsmoothresultvisualization.Thisincludesdatapreprocessing,modelexecution,andvisualization.

4.3

PreparationofDatasets

The eBird observational data on house sparrows gathered between 2015 and 2025, made up the dataset used in this investigation.Importantdetailslikeobservationcounts,latitudeandlongitudecoordinates,andenvironmentalconditions areallincluded.

International Research Journal of Engineering and Technology (IRJET) e-ISSN:2395-0056

Volume: 12 Issue: 04 | Apr 2025 www.irjet.net p-ISSN:2395-0072

Missing values in important variables, such as effort distance, time, and observation counts, are handled effectively to guarantee good data quality. In order to improve the performance of analytical models, feature engineering is used to extractsupplementaryinformationsuchasmonthlyaveragesightings,seasonaltrends,andhabitatclusters.

Inordertoexaminelong-termpatternsinhousesparrowpopulations,timeaggregationisalsocarriedoutbyresampling thedatasetintomonthlyandannual intervals.Thedatasetiswell-structuredandappropriateforgeospatial analysisand predictivemodellingthankstothesepreparationprocedures.

4.4ModelExecution

Several computational models are used in the study to better understand the dynamics of sparrow populations. Timeseries data is analyzed using ARIMA forecasting, which helps detect possible decline periods and forecast future populationpatterns.

K-means clustering is used for spatial clustering, which groups sparrow habitats according to their geographic locations. Thismakesitpossibletodivideareasintohigh-density andlow-densityzones,whichiscrucial forcreatingconservation plansthatwork.

Furthermore, based on geographical and environmental characteristics, observation sites are classified using Random Forest classification into high, moderate, or low sparrow presence. This aids in identifying the areas best suited for sparrowresidenceandconservationinitiatives.

Root Mean Squared Error (RMSE) for ARIMA forecasting, inertia for K-Means clustering, and classification accuracy for Random Forest models are used to assess model performance. These measures guarantee that the models generate accurateandcomprehensibleoutcomes.

4.5IntegratingPredictiveModelingandGeospatialAnalysis

A thorough grasp of house sparrow population patterns is made possible by this study's integration of time-series forecasting, clustering, and classification into a single analytical framework. ARIMA forecasting enables academics to predictpossiblereductionsinvariousplacesbyofferinginsightsintolong-termpopulationtrends.

By graphically depicting high-density and low-density sparrow habitats on satellite maps, K-Means clustering improves geospatial analysis. This method aids in the development of focused conservation strategies in addition to identifying importanthabitats.

Additionally, classification algorithms make it possible to pinpoint the main environmental elements affecting sparrow populations. This study's analysis of observation patterns yields a predictive framework that can direct conservation initiativesandguaranteeearlyactioninregionswheresparrowpopulationsareindanger.

This study provides a data-driven approach to avian conservation and advances our understanding of house sparrow population dynamics by utilizing large-scale ecological information, sophisticated statistical models, and geospatial analysis.

Results & Discussion:

Thestudy'sconclusionsshedimportantlightonthedynamicsofthehousesparrowpopulationbypointingoutpatternsin habitatclustering,populationprojections,andcategorizationoutcomes. Thisstudy offersa thoroughgrasp ofpopulation distribution patterns and the major environmental elements impacting them by combining ARIMA forecasting, K-Means clustering,andclassificationalgorithms.

HouseSparrowPopulationForecast(ARIMAModel)

Time-seriesdataofsparrowsightingswereanalyzed usingtheARIMAmodeltoforecastfuturepopulationchanges.With varyingobservationcounts, thepredictedfindingsshow a slowdrop inhousesparrownumbers. The following forecasts indicatethattherewilllikelybeaconsiderabledecreaseinrecordedsightingsbytheendof2026:

International Research Journal of Engineering and Technology (IRJET) e-ISSN:2395-0056

Volume: 12 Issue: 04 | Apr 2025 www.irjet.net p-ISSN:2395-0072

Withpredictedvaluesrangingbetween242and263observationspermonth,theconfidenceintervalsindicatesignificant unpredictabilityinfuturepopulationpatterns.Althoughthereissomeshort-termstability,thegeneraltendencypointsto along-termdecreasethatcallsforconservationmeasures.

Inkeepingwithnaturaltendencieswheresparrowpopulationspeakduringthebreedingseason(March–June)anddropin thewinter,theforecast'sgraphicrepresentations(arima_forecast.png&population_forecast.png)showcyclicoscillations.

International Research Journal of Engineering and Technology (IRJET) e-ISSN:2395-0056

Volume: 12 Issue: 04 | Apr 2025 www.irjet.net p-ISSN:2395-0072

GeospatialHabitatClustering(K-MeansClustering)

Three separate clusters representing high-, medium-, and low-density zones were created by applying the K-Means clusteringtechniquetohousesparrowhabitats.Thefollowingistheclusterdistribution:

Table 2 : Cluster Distribution

Cluster Density Type Number of Observations

Cluster2 High-Density 36,682

Cluster0 Moderate-Density 26,066

Cluster1 Low-Density 18,980

● Cluster 2 (High density): Foundinruralandsemi-urbanregions,wheretherearemorefoodsourcesandnesting locations.

● Cluster 0 (Moderate-Density): represents suburban areas that exhibit a slow population decline and a mix of urbanbuildingsandgreenspaces.

● Cluster 1 (Low-Density) : correspondslowdensity):correspondstoareasthatare heavilyurbanized,wherethe lack of nesting grounds, habitat degradation, and air pollution have resulted in drastically decreased sparrow populations.

A satellite-based visualization (cluster_distribution_satellite.png) highlights the spatial distribution of sparrow populations across Maharashtra, providing valuable insights for conservation planners to prioritize habitat restoration effortsinlow-densityareas.

International Research Journal of Engineering and Technology (IRJET) e-ISSN:2395-0056

Volume: 12 Issue: 04 | Apr 2025 www.irjet.net p-ISSN:2395-0072

SpeciesClassification&EnvironmentalInfluence(RandomForestModel)

The Random Forest classifier was applied to analyze environmental factors influencing sparrow population distribution. The model achieved an accuracy of 89.6%, demonstrating strong predictive performance. However, feature importance analysisrevealedthatgeospatialcoordinates(latitudeandlongitude)hadnegligibleinfluence,withbothshowingascore of0.0.

This suggests that other unaccounted ecological factors such as vegetation cover, food availability, noise pollution, and air quality may play a more significant role in determining sparrow populations. Future studies should consider integrating additional environmental variables, such as temperature, noise levels, and air pollution indices, to enhance classificationaccuracy.

International Research Journal of Engineering and Technology (IRJET) e-ISSN:2395-0056

Volume: 12 Issue: 04 | Apr 2025 www.irjet.net p-ISSN:2395-0072

According to the ARIMA forecast, the study shows a slow fall in house sparrow populations, especially in metropolitan areas. Even though there are occasional variations, the long-term trend points to a steady decline, highlighting the necessityofconservationmeasures.

Withlow-densityclustersinheavilyurbanizedareasandhigh-densityclustersinruralandsemi-urbanareas,theK-Means clustering analysis clearly shows variances in population density. This demonstrates how urbanization, pollution, and habitatdestructionhaveadetrimentaleffectonsparrownumbers.

Even with its 89.6% accuracy, the Random Forest classification algorithm shows that sparrow population predictions cannotbemadeonlybasedongeospatialcriteria.Thisimpliesthatforimprovedpredictionaccuracy,otherenvironmental factorsincludingfoodavailability,vegetationcover,andairpollutionlevelsshouldbeincluded.

Allthingsconsidered,thestudyshowshowwellAI-drivenecologicalmodellingcananalyzetrendsinbirdpopulations.The results offer important information for conservation planning, determining important habitats, setting restoration priorities, and encouraging the preservation of urban biodiversity. To improve predictions and conservation methods, futureresearchshouldconcentrateongrowingdatasets,addingnewecologicalaspects,andutilizingcutting-edgemachine learningalgorithms.

Conclusion:

Thefallinghousesparrowpopulationandthefactorscausingittodeclinearebetterunderstoodthankstothisstudy.The study shows how to analyze population trends and pinpoint areas most impacted by urbanization using time-series forecasting,habitatgrouping,andgeospatialanalysis.Accordingtothefindings,sparrowsarehavingdifficultyadjustingto denselypopulatedplaceswheretherearefewernestinglocationsandfoodsupplies.Effectiveconservationmeasuresmust be put in place to guarantee population stability, even while AI-driven models provide predictive insights. To improve forecasts and direct conservation efforts, future research should take a more comprehensive ecological approach that integratesavarietyofenvironmentalfactors.

1. TheARIMAmodelrevealsadecliningpopulationtrend,withfluctuationslinkedtoseasonalpatterns.

2. Clustering results indicate higher sparrow densities in rural and semi-urban areas, while urban centres show a sharpdecline.

3. Existing models highlight the importance of integrating additional ecological factors such as temperature, pollution,andvegetationcoverforbetteraccuracy.

4. Conservationmeasureslikeurbangreening,installationofnestingspaces,andreducedpesticideuseareessential forpopulationrecovery.

5. Long-term field studies and real-time monitoring should complement AI-driven predictions to develop effective conservationstrategies.

Future Scope:

1. Including Other Environmental Factors: Toimproveforecastaccuracy,futureresearchshouldincludeelements likefoodavailability,noiselevels,temperaturefluctuations,andairpollution.

2. Improving Model Accuracy: ByintegratingecologicalsimulationswithAI-drivenmethodologies,itispossibleto gainabetterknowledgeofsparrowpopulationdynamicsandhowtheyreacttoenvironmentalchanges.

3. Field-Based Validation: Verifyingforecastsandimprovingconservationtacticswillbemadeeasierbycombining AImodelswithlong-termmonitoringandon-the-groundsurveys.

4. Study Region Expansion: By extending studies outside of Maharashtra to several different regions, a more thoroughgraspofsparrowpopulationpatternsinvarioushabitatswillbepossible.

5. Creation of Conservation Plans: To support sparrow populations in diminishing locations, the results of AIdriven analysis can direct community-driven conservation initiatives, regulatory changes, and urban planning adjustments.

International Research Journal of Engineering and Technology (IRJET) e-ISSN:2395-0056

Volume: 12 Issue: 04 | Apr 2025 www.irjet.net p-ISSN:2395-0072

6. Incorporation of Climate Change Analysis: Analyzinghowclimatechangeisaffectingbirdpopulationsoverthe longrunmightaidinforecastinghazardsandcreatingmitigationstrategies.

7. Public Awareness and Citizen Science Initiatives: Real-timesparrow population monitoringcan befacilitated byinvolvingcommunitiesthroughbirdingprogramsandcrowdsourceddatagathering.

8. Cooperation with Conservation Organizations: Large-scale conservation initiatives can be facilitated by formingpartnershipswithecologicalresearchinstitutesandwildlifeprotectionorganizations.

9. Policy Recommendations for Urban Planning: Makinguseofresearchfindingstosupporttheincorporationof greenspacesandartificialnestingsitesinurbanareas,amongotherbird-friendlycitydesigns.

References/Citations:

1. Sharma, P., Binner, M., & Indian Council of Agricultural Research. (2020). The decline of population of house sparrow in India.InternationalJournalofAgriculturalScience,Volume5.

2. Shansaz, U. H., Jr., Fazili, M. F., Wildlife Research Lab., & Postgraduate Department of Zoology, University of Kashmir. (2022). Distribution and conservation issues of House Sparrow: A review. International Journal of AdvancedScientificResearchandManagement,7(9).

3. Jhajhria,A.&JEZS.(2020). A review on the need for conserving the house sparrow in India.Journal ofEntomology andZoologyStudies,8(2),1157–1159.

4. Paul, M. R. & Consortium & Training Academy for Biosciences (CTAB). (2015). A Review of House Sparrow Population Decline in India.AsiaPacificJournalofResearch,38–40.

5. Modak,B.(2015). Impact of Urbanization on House Sparrow Distribution: A Case Study from Greater Kolkata, India ProceedingsoftheZoologicalSociety,70.

6. Balmori,A.&Hallberg,Ö.(2007). The Urban Decline of the House Sparrow (Passer domesticus): A Possible Link with Electromagnetic Radiation.ElectromagneticBiologyandMedicine,26(2),141-151.

7. Biswal, D. (2021). Urban Growth and the Population of House Sparrows and House Crows: Is the Urban Ecosystem Undergoing a Radical Change? 42(1),17-35.

8. Ali,S.andRipley,R.(1987).CompacthandbookofthebirdsofIndiaandPakistan.OxfordUniversitypress,Delhi, pp:296-314.

9. BirdlifeInternational(2013),“Passerdomesticus,"IUCNRedListofthreatenedspeciesversion2013.2.

10. Dandapat, A., Banerjee, D., & Chakraborty, D. (2010). The case of the Disappearing House Sparrow (Passer domesticusindicus).VeterinaryWorld,3(2),97–100.https://www.cabdirect.org/abstracts/20103054772.html