Professional Documents
Culture Documents
net/publication/360566672
CITATIONS READS
0 93
5 authors, including:
Mayank Sharma
Ambedkar University Delhi
32 PUBLICATIONS 188 CITATIONS
SEE PROFILE
All content following this page was uploaded by Nitin Singh on 09 November 2022.
To cite this article: Venkataraghavan Krishnaswamy, Nitin Singh, Mayank Sharma, Neha Verma &
Amit Verma (2022): Application of CRISP-DM methodology for managing human-wildlife conflicts:
an empirical case study in India, Journal of Environmental Planning and Management, DOI:
10.1080/09640568.2022.2070460
Article views: 56
(Received 15 June 2021; revised 6 April 2022; final version received 11 April 2022)
1. Introduction
Human-Wildlife conflict (HWC) is a negative interaction, either perceived or real,
between humans and wildlife, where one hurts the other (Conover 2001; Decker,
Lauber, and Siemer 2002). HWC is of significant concern to local residents, industry,
governments, and national economies (Heinen and Kattel 1992). Primarily, HWC
occurs due to shared ecological resources that have transformed wildlife habitats into
urban, suburban, and agrarian areas causing degradation, fragmentation, and loss of
wildlife habitat, incapacitation of wildlife corridors, and non-availability of essential
resources in natural habitats (Seeland 2000; Lele and Srinivasan 2013).
HWC may result in direct economic losses and indirectly expends a nation’s finan-
cial and human resources (Barua, Bhagwat, and Jadhav 2013; Anand and
Radhakrishna 2017). There are economic losses due to crop depredation by elephant,
macaque, spotted deer, sambar deer, antelope, and wild boar; livestock depredation by
carnivores such as tiger and leopard; property damage mainly by elephant; and human
injury and loss of life by tiger, bear, and elephant (Heinen and Kattel 1992; Gubbi
et al. 2014; Guru and Das 2021). The abundance of wildlife in their environment may
threaten the local population and farmers (Seeland 2000; Gillingham and Lee 2003;
Rao et al. 2002). It may also result in the retaliatory killing of wildlife (Mateo-Tomas
et al. 2012). HWC undermines the conservation efforts in and around protected areas
and depletes local support for conservation, thus endangering wildlife species (Anand
and Radhakrishna 2017; Hill 2004; Manral et al. 2016). Matseketsa et al. (2019a) ana-
lyze strategies the local population uses to resist conservation efforts and conclude that
it is necessary to provide realistic incentives and actively involve them in conserva-
tion efforts.
In the context of biodiversity, there can be impacts and conflicts, as highlighted by
Young et al. (2010). HWC can emerge from biodiversity impacts (Young et al. 2010).
This can be due to government intervention for people’s livelihoods, policy, or other
reasons. Thus, HWC needs to be managed to maintain the conservation goals, ecosys-
tem, and people’s livelihoods. Protected area managers need to explore new mecha-
nisms and adopt better conservation strategies to lower HWC (Heinen and Kattel
1992). The mechanisms to manage HWC can be mitigative and preventive (Ogra and
Badola 2008). Among mitigative mechanisms, compensation schemes provide relief to
the local population in forested areas affected by HWC (Seeland 2000). While com-
pensation for conflicts is easier to implement than preventive mechanisms (Bautista
et al. 2019), they have mixed results. For example, low payments have no impact on
the intrinsic motivation of stakeholders (Handberg and Angelsen 2019), and positive
results are observed only under suitable conditions (Rao et al. 2002). Sondhi et al.
(2016) found that compensation schemes were not perceived positively by people,
while Pradhan (2018) discussed the inefficiencies and ineffectiveness of the mecha-
nisms delivering compensation. On preventive mechanisms toward HWC, Jamwal,
Tapka, and Parsons (2019) call for conflict-specific strategies. In formulating conserva-
tion strategies, Leverington et al. (2010) and Lu, Kao, and Chao (2012) advocate the
need for methodologies that eliminate subjectivity issues of information and improve
adherence to good governance principles (Lockwood 2010). Loft et al. (2017) opine
that incident analysis could form the basis for designing conservation strategies.
Similar views on data-driven decision-making related to the consumption of natural
resources are echoed by Negi and Maikhuri (2017).
While the HWC literature has acknowledged the need to remove subjectivity in
decision-making, there is a lack of appropriate frameworks or methods to address this
critical issue. Looking at this from a problem-solving paradigm where the large prob-
lem area is HWC, we may adopt problem-solving frameworks to proceed with appro-
priate decision-making. Simon (1960) described four stages-based decision-making:
intelligence, design, choice, and implementation. These stages correspond to stating
the problem, identifying solutions, choosing the best solution, and evaluating the solu-
tion to see its effectiveness. We may achieve the decision-making stages through this
problem-solving paradigm. Simon’s (1960) decision-making framework can be opera-
tionalized through a data-driven approach such as CRISP-DM. CRISP-DM would be
more appropriate in terms of implementation and iteratively identifying solutions to
manage HWC. CRISP-DM methodology provides a structured approach, which can
aid the protected area managers to overcome limitations they might otherwise face
while making decisions. We focus more on this aspect and contextualize CRISP-DM
to protected area management in Section 3. Therefore, the research objectives of this
study are to (a) apply the cross-industry standard process for data mining (CRISP-DM)
Journal of Environmental Planning and Management 3
2. Methods
Addressing the first research objective, we apply CRISP-DM to study the patterns in
HWC. CRISP-DM is an industry-agnostic methodology for extracting knowledge from
data (Wirth and Hipp 2000). CRISP-DM is applied, for instance, in healthcare
(Asamoah and Sharda 2015), insurance (Wang et al. 2019), and new product develop-
ment (Relich 2013). It is more popular than other process models such as Knowledge
Discovery in Databases (KDD) and Sample, Explore, Modify, Model, Assess (SEMMA)
(Shafique and Qaiser 2014). However, to the best of our knowledge, it has not been
adapted or applied in protected areas, which adds to the uniqueness of this work. As
CRISP-DM is usually applied to business environments, we contextualize it by draw-
ing inputs from the relevant frameworks for protected area management (Hockings
et al. 2006; Lockwood 2010).
Addressing the second research objective, we conduct a case study demonstrating
CRISP-DM for HWC management in Ramnagar Forest Division (RFD), India. We
analyze the HWC data and identify geographical hotspots, species-wise hotspots, sea-
sonality, crop loss, livestock loss, and human-loss patterns. Based on our analysis, we
propose conflict-type and site-specific strategies for preventing HWCs. As part of our
study, qualitative interviews validated our analysis and ongoing interventions by pro-
tected area managers. Our second research objective is like Matseketsa et al. (2019b),
who studied strategies adopted by the local population in Save Valley Conservancy
and proposed land-use planning, regular maintenance, and repair of a boundary fence
to reduce HWC.
On the implications, the approach provides an actionable management tool for pro-
tected area managers, based on data-driven decision making, thereby avoiding subject-
ivity and improving rationality, transparency, legitimacy, and accountability toward
good governance (Lockwood 2010). Interviews hint that a participatory approach
involving local stakeholders in planning and implementing preventive strategies for
HWC might improve their effectiveness. We also highlight issues related to data man-
agement and provide suggestions for improvement. We organize the paper as follows.
First, we describe CRISP-DM in detail. Next, we illustrate the step-by-step application
of CRISP-DM in the case study context, culminating in the proposal of preventive
strategies and qualitative assessment. Limitations and conclusions are presented in the
final section.
3. Crisp-DM
This section describes the CRISP-DM stages and contextualizes them to protected area
management. Knowledge discovery is the process of capturing and discovering novel,
valid, valuable, and understandable patterns of information or insights from data
(Relich 2013). Wirth and Hipp (2000) proposed CRISP-DM as an industry and tech-
nology-agnostic reference model for knowledge discovery from data, incorporating the
following stages: (1) business understanding, (2) data understanding, (3) data
4 V. Krishnaswamy et al.
preparation, (4) modeling, (5) evaluation, and (6) deployment. Figure 1 depicts the
stages of CRISP-DM along with the stage-specific activities.
We contextualize CRISP-DM by drawing upon relevant protected area-specific ele-
ments from the Management Effectiveness Evaluation framework (MEE) (Hockings
et al. 2006) and governance principles (Lockwood 2010). Several works have used the
MEE framework or its adapted version, for example, in evaluating marine protected
areas in the UK (Gubbay 2005), Siberia’s protected areas (Anthony and Shestackova
2015), and Taiwan’s protected areas (Lu, Kao, and Chao 2012). While MEE provides
a framework for evaluating protected areas, unlike CRISP-DM, it is a reference model
for evaluation rather than a prescriptive approach for making data-driven decisions.
The authors of the MEE framework acknowledge that “This Best Practice Guideline
(MEE framework) is not intended as a ‘how-to’ manual and does not contain a
detailed methodology but explains the stages in designing and conducting an assess-
ment (Hockings et al. 2006, vii).” Lu, Kao, and Chao (2012, pg. 274) list several prob-
lems, namely (a) absence of in-depth analysis, (b) inability to provide a detailed site or
problem-specific information, and (c) subjectivity of information and over-reliance on
authorities. Lu, Kao, and Chao (2012) and Leverington et al. (2010) provide a direc-
tion for remedying this problem and call for incorporating structured methodologies in
decision-making. We recommend adopting CRISP-DM, contextualized to protected
areas and governance principles as a solution. As shown in Figure 1, contextualization
is done by considering inputs from conservation literature in two stages of CRISP-
DM: business understanding and deployment. We contextualize the business under-
standing step by considering domain-specific inputs from the MEE framework regard-
ing the significance of the environmental problem, the associated values, stakeholders
– wildlife, administrators, and people, supporting legislation and policy instruments.
We contextualize the deployment step by deriving desirable strategies, where
“desirable” is the degree to which they adhere to the principles of good governance,
such as legitimacy, transparency, accountability, fairness, and ethics (Lockwood 2010).
We demonstrate the application of the CRISP-DM through a case study on HWC pre-
vention in the next section.
Journal of Environmental Planning and Management 5
4. Case study
This section provides a stage-wise, step-by-step illustration of how the CRISP-DM
methodology (Figure 1) can be applied in protected areas for HWC management for
deriving preventive strategies.
4.1.1. Background
4.1.1.1. National and regional context [MEE]. RFD lies in the southeastern region of
Uttarakhand, India, and covers an area of 593 km2 (Ahmed et al. 2018). The study
area is divided into four human settlements, from now on referred to as clusters,
namely Haldwani, Kaladhungi, Kotabagh, and Ramnagar, which are either interspersed
or adjoining and contiguous with the RFD. The clusters, except Kotabagh, are adminis-
trative divisions in the district of Nainital, Uttarakhand. About 58.24 km2 of the RFD
is part of a protected space called the Pawalgarh Conservation Reserve (PCR). The
PCR lies partly in Ramnagar and partly in the Kotabagh cluster. The RFD adjoins the
Corbett Tiger Reserve (CTR), which has a high wildlife density on the western side.
The Nandhaur Wildlife Sanctuary flanks the RFD on the eastern border. Figure 2
shows the location of the RFD and the relative location of four clusters.
4.1.1.2. Value [MEE]. India is home to an estimated 2,967 tigers in the wild (Jhala,
Qureshi, and Nayak 2020). Protecting the apex predator is critical to maintaining the
ecosystem balance in and around the protected areas.
4.1.1.3. Significance [MEE]. Minimizing HWC is critical to the socio-economic situ-
ation in the RFD and maintaining a positive perception and favorable attitude toward
conservation attempts at RFD, PCR, and CTR.
4.1.1.4. Stakeholders [MEE]. Stakeholders include humans, wildlife, livestock, crops,
and protected area administration. The dominant wild fauna consists of the Royal
Bengal Tiger, Asian elephant, boars, rhesus monkey, spotted deer, and sambar deer.
Flora includes Sal forests, Sal mixed forests, mixed forests, teak plantations, scrub-
lands, moist riverine forests, dry riverine forests, barren lands, water bodies, and farms
(Corbett 2019; Ahmed et al. 2018). Around 0.58 Mn people reside in the study clus-
ters, with Haldwani being the largest cluster by population (0.36 Mn) (Census 2011a).
From an economic perspective, 61% of the people of Nainital district live in rural
spaces, and farmers and agricultural labor constitute 46% of its working population
(Census 2011b).
4.1.1.5. Legislation and policy [MEE]. Resources for RFD are available and adminis-
tered through the Compensatory Afforestation Fund Act, 2016 (Campa 2018). The
principal objectives of the legislation are (a) Conservation, protection, regeneration,
and management of existing natural forests, (b) Conservation, protection, and manage-
ment of wildlife and its habitat within and outside the protected areas, including
consolidation of the protected areas, (c) Compensatory afforestation. The budget for
2018-19 was USD 45 Mn and includes grants for HWC prevention measures such as
6 V. Krishnaswamy et al.
The village name was not available separately and had to be extracted from
the address.
Inconsistency in the compilation of address field. The village name was either
preceded or succeeded by the name of the nearest major town or post office.
Inconsistency in the language use, with instances of the transliterated name of
the village (from local dialect) instead of the correct name.
Inconsistent formatting of the incident date field.
8 V. Krishnaswamy et al.
The values of the attribute Conflict Animal were singular in some instances and
plural in other cases. This inconsistency was further observed in the attribute
Livestock Lost.
Journal of Environmental Planning and Management 9
4.4. Modeling
4.4.1. Select technique [CRISP-DM]
Descriptive and spatial analyses were employed.
The reason for the variation in leopard conflict between the Kotabagh and Ramnagar
regions is that tigers and leopards are sympatric carnivores with a large overlap in their
diet leading to spatial partitioning between these two species (Lovari et al. 2015). This
results in the displacement of leopards from Ramnagar toward Kotabagh, which is rela-
tively less accessible to tigers (Harihar, Pandav, and Goyal 2011). Elevated levels of
human-elephant conflicts are seen in the Haldwani cluster as it contains corridors of crit-
ical importance to the elephant’s biological requirements (Tiwari et al. 2017).
4.4.3.3. Patterns of crop loss. We analyzed a total number of 288 crop-loss incidents,
with 50% occurring in the Ramnagar cluster, followed by Haldwani (29%), Kotabagh
(15%), and Kaladhungi (6%). Table 2 shows the details and highlights the differences
across the clusters with respect to the conflict animal involved in crop loss. While
Haldwani accounts for fewer incidents than Ramnagar, the damage per incident is 33%
higher in Haldwani. This is probably because elephants almost exclusively cause crop loss
in Haldwani. In Ramnagar and Kotabagh, which are closer to CTR and PCR, the damage
is primarily attributable to boars and other ungulates. In the case of Kaladhungi and
Haldwani, crop loss is primarily due to elephants and boars. Teda and Amgarhi are the
most affected villages, accounting for almost 30% of damages in compensation. Dense for-
ests surround these villages, and the easy availability of crops attracts herbivores.
4.4.3.4. Patterns of livestock loss. A total number of 370 livestock loss incidents
(refer Table 3) were reported with approximately 51% in the Ramnagar cluster, fol-
lowed by Kotabagh (39%), Haldwani (5%), and Kaladhungi (5%). A sum of 76,545
USD was provided as compensation. Tigers, leopards, and snakes are involved in live-
stock loss across clusters. The loss of livestock due to tigers and leopards is compara-
tively higher in Ramnagar and Kotabagh than in Haldwani and Kaladhungi due to
their proximity to the CTR and PCR. Unlike Ramnagar, where tigers are primarily
responsible for livestock loss, tigers and leopards are involved in other clusters.
Perhaps, this is due to the proximity of Ramnagar to the CTR. The implication is that
preventive measures in the Ramnagar cluster should focus more on tigers, while such
measures should be aimed at both tigers and leopards in the other clusters. Teda and
Nathujala are the most affected villages from a hotspot perspective, accounting for
almost 17% of damages in compensation. Teda village, surrounded by dense forests,
tops the crop and livestock loss list. Moreover, villagers venture into forests regularly
for shepherding their livestock, collecting fuelwood and other non-timber forest pro-
duce such as curry leaves, thus experiencing increased incidences of HWC.
Later, during qualitative interviews, it was also found that villagers having more
than a few cattle (3-4 cows) tend to graze their cattle away from their homes in for-
ested patches attracting predators. Cattle owners do accompany their cattle during graz-
ing. However, they reported incidents where predators were bold enough to snatch
cattle right under their supervision. Further, many incidents happen at night when vil-
lagers do not keep vigil.
4.4.3.5. Patterns of human loss. Human loss denotes any type of injury, or a fatality
suffered by a human victim in HWC. The frequency of incidents of human loss is
much lower compared to crop or livestock loss. As shown in Table 4, cluster-wise, of
the 29 human-loss incidents, 12 (41%) were reported in Kotabagh, followed by 11 in
Ramnagar (38%), 5 in Haldwani (17%), and 1 in Kaladhungi (4%). Moreover, it can
be observed that while the incidents are less compared to other clusters, the total com-
pensation awarded is highest in Haldwani as the amount includes compensation for
Table 2. Crop loss incidents and compensation in Ramnagar Forest Division.
Unknown 01 45 45 45
Total 288 6,027 42 2,399 55 4,624 56 1,004 56 14,054
13
14
Conflict Animal Incidents Total Avg. Total Avg. Total Avg. Total Avg. Total
Leopard 104 2,272 126 12,489 192 1,016 92 1,609 176 17,386
Snake 10 430 143 931 186 42 42 42 42 1,447
Tiger 255 37,666 230 16,328 224 1,277 182 2,230 223 57,501
Panther 01 212 212 212
Total 370 40,368 216 29,960 208 2,336 123 3,881 194 76,545
Journal of Environmental Planning and Management 15
two fatalities due to snake bites. Species involved in human loss differ across clusters.
For example, the Kaladhungi cluster is free of any incident involving any species other
than snakes. Incidents involving tigers and elephants are seen in Ramnagar, Haldwani,
and Kotabagh. Instances of conflict with leopards are largely restricted to Kotabagh,
while wild boars are involved in a couple of incidents in Kotabagh and Ramnagar.
Again, we can see that tiger incidents are mainly limited to Ramnagar, while leopard
incidents are more in Kotabagh, emphasizing the association between the conflict and
the density of animals. If we exclude villages of Haldwani where snake bite occur-
rences are reported, Bajauniya-Haldu and Patkote are the most affected, accounting for
almost 25% of damages in compensation.
4.4.3.6. Month-wise patterns. The monthly averages of various HWC incidents,
namely loss of human life, crops, and livestock, are presented in Figure 6. Crop loss
incidents show a monthly pattern distinct from livestock and human loss occurrences.
Furthermore, crop losses peak in March and above average during April and peak
again in August-October. This can be correlated to the region’s agricultural calendar.
For example, the crop ripening time for wheat, which is widely grown in these areas,
starts in February and is harvested by April. The mature crop with the developed
grains attracts herbivores, thus leading to high incidences of crop losses. Maize crops
also start maturing during the same season, attracting elephants and other herbivores.
Similarly, rice is another dominant crop and ripens between August and September,
explaining the second peak (Sukumar 1990).
During the period July–October, incidents of livestock loss peak or are above aver-
age. A similar pattern was reported from the Kanha-Achanakmar Corridor of Central
India (Ahmed et al. 2012). Perhaps, this is because of dense foliage during the mon-
soon and post-monsoon seasons (rainy season). Dense foliage and widespread water
availability cause dispersion of wild animals, and carnivores find livestock easy targets
rather than wild prey. Incidents of human loss are above average during the warmer
months between May and October across all the clusters.
4.5. Evaluation
4.5.1. Evaluate analysis [CRISP-DM]
This step determines whether access to knowledge (patterns) would aid the formulation
of strategies for reducing HWC. We analyzed data and identified hotspots by clusters
and species, along with the patterns of crop loss, livestock loss, human loss, both by
time and space dimensions. The patterns were detailed in Section 4.4.3. These patterns
helped determine the strategies for (a) tigers and leopards and (b) elephants, boars, and
herbivores, as detailed in Section 4.6.2.
4.6. Deployment
4.6.1. Governance principles [MEE]
The strategies proposed in this work (Section 3.6.2) conform to the good governance
principles suggested by Lockwood (2010), as discussed below.
4.6.1.1. Instrumental rationality [MEE]. Data-driven inputs provide the logic for
choosing a specific strategy for preventing HWC. Instead of a single broad-based strat-
egy, species and conflict-specific strategies were adopted based on the patterns (know-
ledge) found.
4.6.1.2. Communicative rationality [MEE]. The structured methodology through which
the data is collected, analyzed, and interpreted improves the quality of patterns mined
from data. In addition, CRISP-DM is transparent and less manipulative, making the strat-
egies traceable to the patterns mined, making the process communicatively rational.
4.6.1.3. Legitimacy [MEE]. From a strategy development standpoint, the strategies
are developed to reduce HWC, which aligns with the objectives of the protected area.
From a data standpoint, the data which forms the inputs for these strategies is from
compensation records, which lends credibility to the exercise.
4.6.1.4. Transparency [MEE]. The data-driven approach makes decision-making trans-
parent and traceable to the data. As mentioned earlier, the strategies are proposed based on
the patterns of crop loss, livestock loss, human loss both by time and space dimensions.
The analysis in Section 4.4.3 and the figures and tables provide a traceable path from the
evidence (summarized data) to the proposed strategies, detailed in Section 4.6.2.
4.6.1.5. Accountability [MEE]. Protected area managers are responsible for their
actions to the higher authorities. Data-driven inputs make the strategies robust and
more effective.
18 V. Krishnaswamy et al.
4.6.1.6. Fairness [MEE]. The source data used to arrive at the patterns and strategies
is from the records of the forest department. It being factual data and the traceable
methodology adopted (MEE þ CRISP-DM) ensure no bias in developing the conflict-
specific strategies (in Section 4.6.2) and making them fair.
4.6.1.7. Ethics [MEE]. The strategies adopted help find the right balance between
human and wildlife needs.
infrastructure and other development projects that fragment rich and critical wildlife
habitats and block vital corridors. Strategies should be aimed at anchoring these wild
animals in the forest through the provision of food and water, that is, by developing
grasslands, improving the habitat, and providing water holes, constructing protective
barriers such as solar fencing around human habitations, and encouraging alternative
cropping patterns in the fringe areas.
In Haldwani, crop loss is more elephant-related, whereas, in Ramnagar, it is more
often due to boars and other ungulates such as spotted deer, sambar, etc. Crop loss pat-
terns also indicate that smaller herbivores venture out less than larger herbivores for
foraging. One observation is that land-locked villages (by forests) are heavily prone to
HWC. Trend analysis indicates an association between the agricultural calendar and
HWC. In crop loss due to wildlife, abatement mechanisms must be implemented
through ring-fencing and other detractive measures. The mitigation strategies for crop-
raiding are species-dependent, such as elephant-proof walls, solar fencing for elephants
and other herbivores, and stone walls for wild boars.
Other mechanisms include incentive programs to support land stewardship by the
landowners themselves. Initiatives proved the efficacy of solar fencing around agricul-
tural fields, resulting in reduced crop damage and substantial financial savings. The
Haldwani cluster mainly requires interventions for reducing crop depredation by ele-
phants. It is also necessary to create awareness and sensitize people living in high-con-
flict zones. Our analysis shows that management strategies must be consistent with the
nature of the conflict. We summarize the cluster and species-specific management
strategies in Table 5, along with relevant references.
The interview findings suggest that the effectiveness of interventions listed in Table
5 can be monitored and localized to the needs of a specific locality. For example, the
interviews show that the solar fence has been effective. At the same time, the design
could be enhanced to prevent Nilgai from jumping over the fence or the wild boars
from crawling under it. In addition, interviews indicate that localized awareness pro-
grams and interventions help reduce HWC. This finding also concurs with Heinen and
Shrivastava (2009), who emphasized participatory approaches and education to increase
awareness to achieve conservation goals. The interviews are the first step toward feed-
back on the outcomes, and over time, it may be strengthened by empirical evidence.
Major
Cluster Nature of HWC HWC Driver Strategic Objective Management Strategy
Ramnagar Livestock, crop, Large Carnivores Reduce contact between Reducing resource dependency of local people (Ahmed
and Kotabagh and human loss such as tigers cattle/humans and et al. 2012)
and leopards large carnivores Monitoring the movement of large carnivores by installing
Elephants and Measures to camera traps (Brooks, Kays, and Hare 2020)
other herbivores discourage movement Creating impression pads (Ghimire 2010)
of animals Installation of solar lights near sensitive spots (Dahal
for foraging et al. 2021)
Awareness programs for people living in hotspots (Larson
et al. 2016)
Alternate income generation activities such as ecotourism
(Carpenter 2022)
Provision and subsidized LPG and community biogas
plants for cooking
Solar fencing around human habitations (Dahal et al. 2021)
Solar fencing forming at the boundary of farmlands and the
forest (Matseketsa et al. 2019b)
Haldwani Crop and Elephants Measures to discourage Anchoring animals in the forest by provisioning food and
and Kaladhungi human loss movement of animals water by developing grasslands and water holes (Sinu and
for foraging Nagarajan 2015)
Solar fencing around human habitations (Dahal et al. 2021)
Alternative cropping patterns in the fringe areas
(Matseketsa et al. 2019b)
Realignment of infrastructure that may block vital corridors
or fragment rich wildlife habitats (Jayadevan et al. 2020)
Journal of Environmental Planning and Management
patrolling and awareness programs for promoting conservation and coexistence. Sinu
and Nagarajan (2015) also report similar findings regarding elephants in the Western
Ghats of India and suggest solutions for coexistence. Some were policy decisions
regarding buffer zone management, night patrolling, and waste management plans in
tourist areas. They advocate the need for a proactive warning system for HWC man-
agement. Our study shows that it is possible to identify hotspots through a data-driven
approach, where the forest administration can focus the resources judiciously.
Jayadevan et al. (2020) proposed a simulation-based methodology for tracking animal
movements and identifying areas for restoration to improve the permeability of large
mammals. Similarly, the CRISP-DM approach can use historical data to find animal
movement patterns to reduce HWC.
The work has implications, as follows. CRISP-DM is a methodology for devising
strategies based on a data-driven approach and could be replicated in other protected
areas. Moving beyond HWC, the methodology may be employed to develop strategies
for revenue management, tourism activities within the reserve, and other conservation
activities. However, one must exercise caution in applying CRISP-DM. While it is
tempting to quickly move into the analysis stage, spending time in the beginning
stages is imperative. Incorrect understanding of protected area requirements may lead
to incompatible data mining goals. This may lead to a futile exercise, requiring rework
and wasting time and effort. Though our study has used descriptive and visual ana-
lytics, as part of CRISP-DM, studies may consider data mining methods such as classi-
fication and clustering may be employed in planning and management studies (Salvati
et al. 2015).
Our study highlights the importance of data management. Data cleaning and trans-
formation accounted for 25% of the overall time required from data collection to ana-
lysis in this case study. This problem is a lack of information systems for managing
data and making productive use of the same. Specifically, in this case, problems arose
due to a lack of data definition, which, in turn, led to non-standardized ways of col-
lecting and scribing data. Consequently, data had missing values and varied in units,
language, and format. All this required significant data cleansing efforts. The critical
data, namely the latitude and longitude coordinates, are measured at the village level.
While this is the closest one can get manually, a better approach would be to acquire
the latitude and longitude coordinates using a mobile phone at the incident spot. An
even better approach would be to scribe the data into a geo-enabled mobile applica-
tion, which, by design, would enable the recording officer to select data values instead
of scribing them on paper or a spreadsheet.
Most of the successful conservation outcomes depend on the ability of the pro-
tected area management to compensate for HWC damages in the short term and pre-
vent the causes of conflict in the long term. Every prevention measure has a cost
attached to it. Besides, resources such as budgets and the staff available with the man-
ager to implement the prevention measures are always limited. The findings and ana-
lysis of this study provide managers with clarity on the type of conflict, its spatial
distribution, and its intensity. Hence, the study offers an actionable management tool
to create prevention strategies for reducing HWC by focusing on specific interventions
related to conflict and location, thereby optimizing resource use while maximizing the
effectiveness of interventions. Basing the strategy on a data-driven approach renders
the entire management process more legitimate, transparent, accountable, and fair,
thereby adhering to the principles of good governance.
24 V. Krishnaswamy et al.
Our work has the following limitations. First, the data collected did not contain
attributes related to livestock ownership, limiting our ability to analyze any correlation
between the number of incidents and livestock ownership. Second, it did not include
information about existing preventive measures implemented in the incident locations.
As such, some hot spots, which are otherwise vulnerable, would not reveal themselves
in the analysis. We recommend that the type and number of preventive strategies
employed at a village level be considered in future studies of this nature. Third, while
we have identified preventive strategies, their outcomes were validated to a certain
extent through interviews and not empirically. Empirical validation of the implemented
strategies and the local population’s perception of conservation efforts could be part of
future studies. Fourth, in our research, we worked on the compensation data for a pro-
tected area. The approach in our work could be replicated in other HWC data. Finally,
the CRISP-DM methodology could be replicated in other protected areas.
Acknowledgements
We would like to thank anonymous reviewers for their suggestions and advice. We would also
like to acknowledge the support from the officials of the Department of Forest, Uttarakhand
(India), and residents of the area covered in this study.
Disclosure statement
No potential conflict of interest was reported by the authors.
Funding
This research did not receive any funding.
ORCID
Venkataraghavan Krishnaswamy http://orcid.org/0000-0001-6862-9583
Nitin Singh http://orcid.org/0000-0002-2466-0252
Mayank Sharma http://orcid.org/0000-0002-5719-0015
References
Ahmed, R. A., K. Prusty, J. Jena, C. Dave, S. K. Das, H. K. Sahu, and S. D. Rout. 2012.
“Prevailing Human Carnivore Conflict in Kanha-Achanakmar Corridor.” Central India.
World Journal of Zoology 7 (2): 158–164.
Ahmed, T. H. S. Bargali, N. Verma, and A. Khan. 2018. “Status of Wildlife Habitats in
Ramnagar Forest Division, Terai-Arc Landscape, Uttarakhand, India.” Geoscience Research
3(1), 1–8.
Anand, S., and S. Radhakrishna. 2017. “Investigating Trends in Human-Wildlife Conflict: Is
Conflict Escalation Real or Imagined?” Journal of Asia-Pacific Biodiversity 10 (2):
154–161. doi:10.1016/j.japb.2017.02.003.
Journal of Environmental Planning and Management 25
Anthony, B. P., and E. Shestackova. 2015. “Do Global Indicators of Protected Area
Management Effectiveness Make Sense? A Case Study from Siberia.” Environmental
Management 56 (1): 176–192. doi:10.1007/s00267-015-0495-z.
Asamoah, D., and R. Sharda. 2015. “Adapting CRISP-DM Process for Social Network
Analytics: Application to Healthcare.” In Proceedings of the Twenty-First Americas
Conference on Information Systems, 485–487. Red Hook, NY: Curran Associates, Inc.
Barua, M., S. A. Bhagwat, and S. Jadhav. 2013. “The Hidden Dimensions of Human–Wildlife
Conflict: Health Impacts, Opportunity, and Transaction Costs.” Biological Conservation
157: 309–316. doi:10.1016/j.biocon.2012.07.014.
Bautista, Carlos, Eloy Revilla, Javier Naves, J€org Albrecht, Nestor Fernandez, Agnieszka
Olszanska, Michal Adamec, et al. 2019. “Large Carnivore Damage in Europe: Analysis of
Compensation and Prevention Programs.” Biological Conservation 235: 308–316. doi:10.
1016/j.biocon.2019.04.019.
Brooks, J., R. Kays, and B. Hare. 2020. “Coyotes Living Near Cities Are Bolder: Implications
for Dog Evolution and Human-Wildlife Conflict.” Behaviour 157 (3–4): 289–313. doi:10.
1163/1568539X-bja10002.
Campa. 2018. “Compensatory Afforestation Fund Act.” Accessed 13 June 2021. http://www.
ukcampa.org.in/history
Carpenter, S. 2022. “Exploring the Impact of Climate Change on the Future of Community-
Based Wildlife Conservation.” Conservation Science and Practice 4 (1): e585. doi:10.1111/
csp2.585.
Census. 2011a. “Census of India: Towns and Villages.” Accessed 13 June 2021. http://
censusindia.gov.in/2011census/Listofvillagesandtowns.aspx
Census. 2011b. “Census of India: Tehsils.” Accessed 13 June 2021. http://censusindia.gov.in/
2011census/dchb/0511_PART_B_DCHB_NAINITAL.pdf
Conover, M. R. 2001. Resolving Human-Wildlife Conflicts: The Science of Wildlife Damage
Management. Boca Raton, FL: CRC Press.
Corbett. 2019. “Corbett National Park.” Accessed 13 June 2021. https://www.corbettonline.uk.
gov.in
Dahal, N. K., K. Harada, A. Sudha, R. P. Sapkota, and S. Kandel. 2021. “Impact of Wildlife on
Food Crops and Approaches to Reducing Human Wildlife Conflict in the Protected
Landscapes of Eastern Nepal.” Human Dimensions of Wildlife 1–17. doi:10.1080/10871209.
2021.1926601.
Decker, D. J., T. B. Lauber, and W. F. Siemer. 2002. Human–Wildlife Conflict Management: A
Practitioner’s Guide. Ithaca, NY: Northeast Wildlife Damage Management Research and
Outreach Cooperative.
Ghimire, Y. 2010. Status of Small Carnivores with a Special Focus on Clouded Leopard in
Makalu-Barun National Park. Kathmandu, Nepal: WWF Nepal.
Gillingham, S., and P. C. Lee. 2003. “People and Protected Areas: A Study of Local
Perceptions of Wildlife Crop-Damage Conflict in an Area Bordering the Selous Game
Reserve, Tanzania.” Oryx 37 (03): 316–325. doi:10.1017/S0030605303000577.
Gubbay, S. 2005. “Evaluating the Management Effectiveness of Marine Protected Areas: Using
UK Sites and the UK MPA Programme to Illustrate Different Approaches.” A Report for
WWF (World Wildlife Fund)-UK. http://assets.wwf.org.uk/downloads/mpa_mgmteff0705.pdf
Gubbi, S., M. H. Swaminath, H. C. Poornesha, R. Bhat, and R. Raghunath. 2014. “An
Elephantine Challenge: Human–Elephant Conflict Distribution in the Largest Asian Elephant
Population, Southern India.” Biodiversity and Conservation 23 (3): 633–647. doi:10.1007/
s10531-014-0621-x.
Guru, B. K., and A. Das. 2021. “Cost of Human-Elephant Conflict and Perceptions of
Compensation: Evidence from Odisha, India.” Journal of Environmental Planning and
Management 64 (10): 1770–1794. doi:10.1080/09640568.2020.1838264.
Handberg, Ø. N., and A. Angelsen. 2019. “Pay Little, Get Little; Pay More, Get a Little More:
A Framed Forest Experiment in Tanzania.” Ecological Economics 156: 454–467. doi:10.
1016/j.ecolecon.2016.09.025.
Harihar, A., B. Pandav, and S. P. Goyal. 2011. “Responses of Leopard Panthera Pardus to the
Recovery of a Tiger Panthera Tigris Population.” Journal of Applied Ecology 48 (3):
806–814. doi:10.1111/j.1365-2664.2011.01981.x.
26 V. Krishnaswamy et al.
Heinen, J. T., and B. Kattel. 1992. “Parks, People, and Conservation: A Review of Management
Issues in Nepal’s Protected Areas.” Population and Environment 14 (1): 49–84. doi:10.
1007/BF01254607.
Heinen, J. T., and R. J. Shrivastava. 2009. “An Analysis of Conservation Attitudes and Awareness
around Kaziranga National Park, Assam, India: Implications for Conservation and
Development.” Population and Environment 30 (6): 261–274. doi:10.1007/s11111-009-0086-0.
Hill, C. M. 2004. “Farmers’ Perspectives of Conflict at the Wildlife–Agriculture Boundary:
Some Lessons Learned from African Subsistence Farmers.” Human Dimensions of Wildlife
9 (4): 279–286. doi:10.1080/10871200490505710.
Hockings, M., S. Stolton, F. Leverington, N. Dudley, and J. Courrau. 2006. Evaluating
Effectiveness: A Framework for Assessing Management Effectiveness of Protected Areas.
Cambridge, UK: IUCN.
Jamwal, P. S., J. Takpa, and M. H. Parsons. 2019. “Factors Contributing to a Striking Shift in
Human–Wildlife Dynamics in Hemis National Park, India: 22 Years of Reported Snow
Leopard Depredation.” Oryx 53 (1): 58–62. doi:10.1017/S0030605317001892.
Jayadevan, A., R. Nayak, K. K. Karanth, J. Krishnaswamy, R. DeFries, K. U. Karanth, and S.
Vaidyanathan. 2020. “Navigating Paved Paradise: Evaluating Landscape Permeability to
Movement for Large Mammals in Two Conservation Priority Landscapes in India.”
Biological Conservation 247: 108613. doi:10.1016/j.biocon.2020.108613.
Jhala, Y. V., Q. Qureshi, and A. K. Nayak. 2020. Status of Tigers, Co-Predators and Prey in
India 2018. Dehradun: National Tiger Conservation Authority, Government of India, New
Delhi and Wildlife Institute of India. Accessed 13 June 2021. http://moef.gov.in/wp-content/
uploads/2020/07/Tiger-Status-Report-2018_For-Web_compressed_compressed.pdf
Larson, L. R., A. L. Conway, S. M. Hernandez, and J. P. Carroll. 2016. “Human-Wildlife
Conflict, Conservation Attitudes, and a Potential Role for Citizen Science in Sierra Leone,
Africa.” Conservation and Society 14 (3): 205–217. doi:10.4103/0972-4923.191159.
Lele, S., and V. Srinivasan. 2013. “Disaggregated Economic Impact Analysis Incorporating
Ecological and Social Trade-Offs and Techno-Institutional Context: A Case from the Western
Ghats of India.” Ecological Economics 91: 98–112. doi:10.1016/j.ecolecon.2013.03.023.
Leverington, F., K. L. Costa, H. Pavese, A. Lisle, and M. Hockings. 2010. “A Global Analysis of
Protected Area Management Effectiveness.” Environmental Management 46 (5): 685–698.
Lockwood, M. 2010. “Good Governance for Terrestrial Protected Areas: A Framework, Principles
and Performance Outcomes.” Journal of Environmental Management 91 (3): 754–766.
Loft, L., D. N. Le, T. T. Pham, A. L. Yang, J. S. Tjajadi, and G. Y. Wong. 2017. “Whose
Equity Matters? National to Local Equity Perceptions in Vietnam’s Payments for Forest
Ecosystem Services Scheme.” Ecological Economics 135: 164–175. doi:10.1016/j.ecolecon.
2017.01.016.
Lovari, S., C. P. Pokheral, S. R. Jnawali, L. Fusani, and F. Ferretti. 2015. “Coexistence of the
Tiger and the Common Leopard in a Prey-Rich Area: The Role of Prey Partitioning.”
Journal of Zoology 295 (2): 122–131. doi:10.1111/jzo.12192.
Lu, D. J., C. W. Kao, and C. L. Chao. 2012. “Evaluating the Management Effectiveness of Five
Protected Areas in Taiwan Using WWF’s RAPPAM.” Environmental Management 50 (2):
272–282.
Manral, U., S. Sengupta, S. A. Hussain, S. Rana, and R. Badola. 2016. “Human Wildlife
Conflict in India: A Review of Economic Implication of Loss and Preventive Measures.”
Indian Forester 142 (10): 928–940.
Mateo-Tomas, P., P. P. Olea, I. S. Sanchez-Barbudo, and R. Mateo. 2012. “Alleviating
Human–Wildlife Conflicts: Identifying the Causes and Mapping the Risk of Illegal
Poisoning of Wild Fauna.” Journal of Applied Ecology 49 (2): 376–385. doi:10.1111/j.1365-
2664.2012.02119.x.
Matseketsa, G., B. B. Mukamuri, N. Muboko, and E. Gandiwa. 2019a. “An Assessment of
Local People’s Support to Private Wildlife Conservation: A Case of Save Valley
Conservancy and Fringe Communities, Zimbabwe.” Scientifica 2019: 1–11. doi:10.1155/
2019/2534614.
Matseketsa, G., N. Muboko, E. Gandiwa, D. M. Kombora, and G. Chibememe. 2019b. “An
Assessment of Human-Wildlife Conflicts in Local Communities Bordering the Western Part
of save Valley Conservancy, Zimbabwe.” Global Ecology and Conservation 20: E 00737.
doi:10.1016/j.gecco.2019.e00737.
Journal of Environmental Planning and Management 27
Negi, V. S., and R. K. Maikhuri. 2017. “Forest Resources Consumption Pattern in Govind
Wildlife Sanctuary, Western Himalaya, India.” Journal of Environmental Planning and
Management 60 (7): 1235–1252. doi:10.1080/09640568.2016.1213707.
Nyhus, P. J. 2016. “Human–Wildlife Conflict and Coexistence.” Annual Review of Environment
and Resources 41 (1): 143–171. doi:10.1146/annurev-environ-110615-085634.
Ogra, M., and R. Badola. 2008. “Compensating Human–Wildlife Conflict in Protected Area
Communities: Ground-Level Perspectives from Uttarakhand.” Human Ecology 36 (5):
717–729. doi:10.1007/s10745-008-9189-y.
Pradhan, V. 2018. Developing Strategies to Mitigate Human Wildlife Conflict in the Sikkim
Himalayas, India. Gangtok, India: Ashoka Trust for Research in Ecology and the
Environment and London: The Rufford Foundation. https://ruffordorg.s3.amazonaws.com/
media/project_reports/20218-1%20Detailed%20Final%20Report.pdf
Rao, K. S., R. K. Maikhuri, S. Nautiyal, and K. G. Saxena. 2002. “Crop Damage and Livestock
Depredation by Wildlife: A Case Study from Nanda Devi Biosphere Reserve.” Journal of
Environmental Management 66 (3): 317–327.
Relich, M. 2013. “Knowledge Discovery from an ERP Database in the Context of New Product
Development.” Informatyka Ekonomiczna 2 (28): 100–111.
Salvati, L., C. Kosmas, O. Kairis, C. Karavitis, S. Acikalin, A. Belgacem, A. Sole-Benet., et al.
2015. “Unveiling Soil Degradation and Desertification Risk in the Mediterranean Basin: A
Data Mining Analysis of the Relationships between Biophysical and Socioeconomic Factors
in Agro-Forest Landscapes.” Journal of Environmental Planning and Management 58 (10):
1789–1803. doi:10.1080/09640568.2014.958609.
Simon, H. A. 1960. The New Science of Management Decision. New York: Harper & Row.
Seeland, K. 2000. “National Park Policy and Wildlife Problems in Nepal and Bhutan.”
Population and Environment 22 (1): 43–62. doi:10.1023/A:1006629531450.
Shafique, U., and H. Qaiser. 2014. “A Comparative Study of Data Mining Process Models
(KDD, CRISP-DM and SEMMA).” International Journal of Innovation and Scientific
Research 12 (1): 217–222.
Sondhi, S. V. Athreya, A. Sondhi, A. Prasad, A. Verma, and N. Verma. 2016. Human Attacks
by Leopards in Uttarakhand, India: An Assessment Based on Perceptions of Affected People
and Stakeholders. A technical report submitted to the Uttarakhand Forest Department. http://
www.conservationindia.org/wp-content/files_mf/Uttarakhand-Leopard-social-survey-reportout-
Apr-2016-Final.pdf.
Sinu, P. A., and M. Nagarajan. 2015. “MEETING REPORT: Human–Wildlife Conflict or
Coexistence: What Do We Want?” Current Science 108 (6): 1036–1038.
Sukumar, R. 1990. “Ecology of the Asian Elephant in Southern India. II. Feeding Habits and
Crop Raiding Patterns.” Journal of Tropical Ecology 6 (1): 33–53. doi:10.1017/
S0266467400004004.
Tableau. 2020. “Which Chart or Graph Is Right for You?.” Accessed 13 June 2021. https://
www.tableau.com/learn/whitepapers/which-chart-or-graph-is-right-for-you
Tiwari, K. S., K. S. Behera, K. Ramkumar, K. C. Sar, D. Swain, and R. Sukumar. 2017.
“Elephant Corridors of North Western India.” In Right of Passage: Elephant Corridors of
India, edited by V. Menon, S. K. Tiwari, K. Ramkumar, S. Kyarong, U. Ganguly, and R.
Sukumar, 424–573. New Delhi: Conservation Reference Series. Wildlife Trust of India.
Wang, C. S., S. L. Lin, T. H. Chou, and B. Y. Li. 2019. “An Integrated Data Analytics Process
to Optimize Data Governance of Non-Profit Organization.” Computers in Human Behavior
101: 495–505. doi:10.1016/j.chb.2018.10.015.
Wirth, R., and J. Hipp. 2000. “CRISP-DM: Towards a Standard Process Model for Data
Mining.” In Proceedings of the 4th International Conference on the Practical Applications
of Knowledge Discovery and Data Mining, Vol. 1, edited by Rakesh Agrawal and Paul
Stolorz, London: Springer-Verlag, April.
Young, J. C., M. Marzano, R. M. White, D. I. McCracken, M. Redpath, D. N. Carss, C. P.
Quine, and A. D. Watt. 2010. “The Emergence of Biodiversity Conflicts from Biodiversity
Impacts: Characteristics and Management Strategies.” Biodiversity and Conservation 19
(14): 3973–3990. doi:10.1007/s10531-010-9941-7.
Zikmund, W. G., J. C. Carr, and M. Griffin. 2013. Business Research Methods. Boston, MA:
Cengage Learning.