NAME: PARONDO, Jerle Mae E. YR/CRS/SEC: 3-BSCE-A SUBJECT: UT 1 (OUTLINE) THESIS ADVISER: Engr.

Kathleen Alucilja

REVIEW OF RELATED LITERATURE

TITLE: Enhancing Road Safety: Black Spot Identification and Accident Prediction Model on Kabacan, North, Cotabato
GENERAL OBJECTIVE:
SPECIFIC OBJECTIVES:
1.

2.

3.

I. VARIABLE (independent or dependent)

# Author, Year Methodology Advantages (Summary, critical evaluation from the study and Disadvantages (loopholes)
Published, & Title of key findings that could contribute on your study)
the Study
Maen Ghadi, 2019, 1. Research Design: The study adopts 1. Improved Accuracy in Identifying Black Spots: The 1. Limited Scope of BSID Methods: While the study
A comparative a comparative research design to study demonstrates that the appropriate combination evaluates four commonly applied BSID methods,
analysis of black examine the effect of
spot identification methodological diversity of road of black spot identification (BSID) methods and there may be other emerging techniques or variations
methods and road network segmentation on the segmentation methods leads to a significant of existing methods that are not considered. This
accident performance of different black
segmentation spot identification (BSID) reduction in false positive and false negative cases. could potentially limit the comprehensiveness of the
methods methods. This indicates that the segmentation method plays a findings and overlook alternative approaches that may
2. Data Collection: The study utilizes
crucial role in enhancing the accuracy of identifying offer improved performance.
road safety-related data, including
accident records, road high-risk locations or black spots on roads. 2. Dependency on Data Quality: The accuracy and
characteristics, and traffic volume
2. Enhanced Performance of Empirical Bayesian reliability of the results are contingent upon the
information, obtained from
relevant transportation authorities Method: The research highlights that the Empirical quality and completeness of the underlying data used
or databases. The data are Bayesian (EB) method outperforms other BSID for analysis. Issues such as underreporting or
collected for the study area,
ensuring representativeness and methods across all segmentation approaches. This misclassification of road accidents could introduce
relevance to the research finding underscores the effectiveness of the EB biases and inaccuracies in the evaluation of BSID and
objectives.
3. BSID Methods Selection: Four method in accurately identifying black spots and segmentation methods.
commonly applied BSID methods prioritizing safety interventions. 3. Generalizability of Findings: The study's findings
are selected for evaluation, namely
3. Methodological Diversity: By evaluating four may be influenced by the specific characteristics of
 empirical Bayesian (EB), excess commonly applied BSID methods against four the study area and the dataset used. As such, the
EB, accident frequency, and different segmentation methods, the study provides generalizability of the results to different geographic
accident ratio. These methods are
chosen based on their prevalence valuable insights into the impact of methodological regions or traffic conditions may be limited, and
in road safety analysis literature diversity on the performance of road safety analysis caution should be exercised when extrapolating
and their suitability for identifying
high-risk locations. techniques. This comprehensive approach allows findings to other contexts.
4. Segmentation Methods Selection: researchers and practitioners to identify the most
Four different segmentation
effective combination of methods for their specific
methods are considered to
organize the road network into context.
homogeneous segments. These
4. Contribution to Safety Performance Function (SPF)
include spatial clustering, constant
length, constant traffic volume, Development: The evaluation of segmentation
and the standard Highway Safety methods based on the accuracy of the developed
Manual segmentation method.
Each method is chosen for its safety performance function (SPF) contributes to the
distinct approach to segmenting advancement of road safety analysis methodologies.
road data.
5. Evaluation Criteria: Two This helps in developing more reliable models for
evaluation criteria are employed to predicting and mitigating road safety risks.
assess the performance of the
BSID and segmentation methods.
The first criterion involves
evaluating the accuracy of the
developed safety performance
function (SPF) resulting from each
segmentation method. The second
criterion applies consistency tests
to compare the joint performances
of the BSID and segmentation
methods.
6. Statistical Analysis: Statistical
techniques such as regression
analysis, hypothesis testing, and
performance metrics computation
are utilized to analyze the data and
evaluate the effectiveness of the
BSID and segmentation methods.
These analyses help identify
trends, patterns, and significant
differences in performance across
 different methodological
approaches.
7. Comparative Analysis: The study
conducts a comparative analysis of
the BSID methods' performance
across the different segmentation
approaches. This analysis involves
examining the impact of
methodological diversity on the
accuracy, reliability, and
effectiveness of each BSID
method in identifying black spots.
8. Conclusion and
Recommendations: Based on the
findings from the comparative
analysis, the study draws
conclusions regarding the relative
performance of the BSID methods
and segmentation approaches.
Recommendations are provided for
selecting appropriate
methodological combinations to
enhance the accuracy and
efficiency of road safety analysis
in practice.

2. Jin Zhang, Xiao-bo 1. Research Design: The study adopts 1. Enhanced Accuracy: The proposed BSID model 1. Limited Scope of Study Sites: The study evaluates the
Qu & Ya-dan Yan, a comparative research design to addresses a critical limitation in traditional proposed BSID model based on a high-speed
2018, A evaluate the effectiveness of the
Comparative Study proposed improved black spot techniques by considering the impact of seasonality, motorway and an urban arterials road, which may not
of Weather Effects identification (BSID) model in specifically wet/dry crash rate ratio. By fully represent the diversity of road conditions and
on Black Spot accounting for the impact of
Identification for seasonality, particularly wet incorporating this factor, the model provides a more environments encountered in real-world settings. The
Motorways and surface conditions, on crash rates. accurate assessment of crash risk, particularly in findings may therefore lack generalizability to other
Urban Arterials 2. Data Collection: Road crash data,
areas prone to weather-related accidents. types of roads and regions.
Roads including crash locations, severity,
and weather conditions (dry or 2. Economical Resource Allocation: By differentiating 2. Data Availability and Quality: The effectiveness of
rainy), are collected from relevant
crashes based on surface conditions (wet or dry the proposed model relies heavily on the availability
transportation authorities or
databases for both a high-speed roads), the model enables more targeted and and quality of crash data, as well as data on road
motorway and a relatively low- economical allocation of resources for road safety surface conditions. Limited or incomplete data may
speed urban arterials road. The
 data are collected over a specified measures. This allows authorities to prioritize introduce biases or inaccuracies in the analysis,
time period to capture variations in interventions such as signage and resurfacing where potentially undermining the reliability of the model's
crash rates across different
seasons. they are most needed, optimizing the use of available findings.
3. Development of Improved BSID resources. 3. Methodological Complexity: While the proposed
Model: The proposed BSID model
incorporates the wet/dry crash rate 3. Applicability Across Road Types: The comparative BSID model offers improvements over traditional
ratio to differentiate crashes study conducted on both high-speed motorways and techniques, its implementation may require
occurring due to poor surface
low-speed urban arterials demonstrates the specialized expertise and resources for data
conditions, particularly wet roads.
The model is developed based on versatility and applicability of the proposed model collection, analysis, and interpretation. This could
statistical analysis and regression
across different road types. This suggests that the pose challenges for smaller agencies or jurisdictions
techniques, considering variables
such as crash severity, road model can be effectively utilized in various settings, with limited technical capacity.
characteristics, and weather enhancing its practical utility for road safety 4. External Validity: While the model demonstrates
conditions.
4. Comparative Study: The study management. good applicability in the study context, its
conducts a comparative analysis 4. Minimization of Rainfall Impact: The study findings performance in other geographic locations or under
between the traditional BSID
techniques and the proposed indicate that the proposed BSID model successfully different climatic conditions remains uncertain.
improved BSID model. This minimizes the impact of rainfall on the results of Further validation studies in diverse settings are
analysis involves applying both
BSID analysis. This is a significant advantage as it needed to establish the external validity and
methods to the collected crash data
from the high-speed motorway and ensures that the identified black spots are not robustness of the proposed model.
urban arterials road to assess their disproportionately influenced by weather conditions,
effectiveness in identifying black
spots and accounting for the leading to more robust and reliable outcomes.
impact of rainfall.
5. Evaluation Criteria: The evaluation
of the BSID methods is based on
several criteria, including
accuracy, sensitivity, specificity,
and predictive performance. These
criteria are used to assess the
ability of each method to
differentiate crashes caused by
poor surface conditions and to
minimize the impact of rainfall on
the results of the BSID analysis.
6. Statistical Analysis: Statistical
techniques such as regression
analysis, hypothesis testing, and
 performance metrics computation
are employed to analyze the crash
data and evaluate the performance
of the BSID methods. These
analyses help identify significant
differences in the effectiveness of
the traditional and improved BSID
models in accounting for
seasonality effects.
7. Validation: The validity of the
proposed improved BSID model is
assessed through validation
techniques, such as cross-
validation or comparison with
historical crash data. This ensures
that the model accurately predicts
black spots and effectively
accounts for the impact of wet
surface conditions on crash rates.
8. Conclusion and
Recommendations: Based on the
findings from the comparative
study, the study draws conclusions
regarding the applicability and
effectiveness of the proposed
improved BSID model.
Recommendations are provided for
its implementation in practice to
enhance road safety planning and
resource allocation strategies,
particularly in areas prone to wet
surface conditions.

3. Zhanyong Fan, 1. Data Collection: The study utilizes 1. Comprehensive Analysis of Influencing Factors: The 1. Data Quality and Availability: The effectiveness of
Chun Liu, Dongjian traffic accident data from Suzhou study conducts a fusion analysis of multi-source the proposed black spot identification and accident
Cai & Shun Yue Industrial Park, capturing various
2019, Research on factors such as time, location, influencing factors involved in traffic accident black prediction model relies heavily on the quality and
black spot people, roads, vehicles, and spots, going beyond traditional approaches that only availability of traffic accident data. Issues such as
identification of environmental conditions. This
safety in urban multi-source data forms the basis consider time and space factors. This comprehensive underreporting, data inconsistency, and data
traffic accidents for the analysis of traffic accident analysis allows for a more holistic understanding of incompleteness may introduce biases and
based on machine black spots.
inaccuracies, potentially undermining the reliability
learning method 2. Fusion Analysis: A fusion analysis the factors contributing to black spot formation. of the model's predictions.
approach is employed to integrate 2. Improved Accuracy of Black Spot Identification: By
multiple influencing factors
involved in traffic accident black utilizing a support vector machine (SVM) method 2. Model Complexity and Implementation Challenges:
spots. This involves synthesizing and a deep neural network (DNN) algorithm, the The use of advanced machine learning techniques
data from different sources to
identify patterns and correlations study enhances the accuracy of black spot such as support vector machines and deep neural
that contribute to the concentration identification. The SVM method maximizes the networks may introduce complexity and
of accidents in specific locations.
classification interval, while the DNN algorithm implementation challenges. Building and training
3. Support Vector Machine (SVM)
Method: The study utilizes a leverages relevant data category information to these models require specialized expertise and
Support Vector Machine (SVM)
identify accident black spots more effectively. computational resources, which may not be readily
method based on maximizing the
classification interval to train a 3. Feature-Based Identification Method: The proposed available in all contexts.
complex model for optimal feature-based black spot identification method based
learning of accident black spots.
SVM is chosen for its ability to on deep neural network offers a novel approach to 3. Generalizability to Other Contexts: While the study
handle complex, high-dimensional identifying accident-prone areas. By incorporating demonstrates promising results in the context of
data and its effectiveness in
classification tasks. various features and data categories, the model can Suzhou Industrial Park, the generalizability of the
4. Deep Neural Network (DNN) capture complex patterns and relationships, leading proposed model to other geographic locations and
Algorithm: To address the rapid
to improved accuracy and prediction capabilities. urban environments remains uncertain. The
growth of traffic accident multi-
source data, a black spot 4. Dynamic Adaptive Machine Learning Architecture: effectiveness of the model may vary depending on the
identification algorithm based on The study introduces a dynamic adaptive machine unique characteristics and dynamics of different
deep neural networks (DNN) is
proposed. This involves learning architecture, which allows the model to regions.
establishing a DNN architecture to adapt and evolve over time in response to changes in
analyze relevant data categories
and improve the accuracy of black traffic patterns and accident data. This adaptive 4. Interpretability of Results: The complex nature of the
spot identification. approach ensures the continued relevance and machine learning algorithms used in the study may
5. Feature-Based Identification: A
effectiveness of the black spot identification model limit the interpretability of the model's results.
feature-based black spot
identification method is proposed, in dynamic urban environments. Understanding the underlying factors driving black
leveraging the capabilities of deep spot identification and accident prediction may be
neural networks to extract
meaningful features from the data. challenging, particularly for stakeholders without
This approach aims to enhance the expertise in machine learning.
effectiveness of black spot
identification by focusing on
relevant data attributes.
6. Dynamic Adaptive Machine
 Learning Architecture: The study
develops a dynamic adaptive
machine learning architecture to
accommodate changes in the data
environment over time. This
architecture enables continuous
learning and adaptation of the
black spot identification model to
evolving traffic conditions and
patterns.

4. Salvatore Cafiso & 1. Definition of Black Spots: The 1. Quantitative Precision Assessment: The study 1. Limited Scope: The study focuses specifically on
Giacomo Di study recognizes that defining a addresses the challenge of quantitatively defining the two-lane rural roads with low to medium traffic
Silvestro, 2011, black spot is not straightforward
Performance of and that high crash counts may not precision of estimates obtained through different volumes. While this allows for a detailed analysis
Safety Indicators always indicate a genuine safety procedures for identifying black spots. By using within this context, the findings may not be directly
in Identification of problem. This suggests a nuanced
Black Spots on understanding of black spot Monte Carlo simulation to generate theoretical crash applicable to other types of roadways or traffic
Two-Lane Rural identification, considering factors data, the research provides a method for objectively conditions. Therefore, the generalizability of the
Roads such as traffic volume, segment
assessing the accuracy and efficiency of different study's conclusions may be limited.
length, and crash observation
period. black spot identification procedures. 2. Simulation-Based Analysis: While the use of Monte
2. Evaluation of Safety Indicators:
2. Comparison of Procedures: The study compares Carlo simulation provides a valuable tool for
Various safety indicators are
considered, including observed the accuracy and efficiency of various procedures for assessing the accuracy of black spot identification
frequency of crashes, crash rate, identifying black spots, including observed crash procedures, the reliance on theoretical crash data may
empirical Bayesian (EB)
estimation, and potential for safety frequency, crash rate, empirical Bayesian estimation, not fully capture the complexity of real-world crash
improvement (PSI). These and potential for safety improvement (PSI). This patterns. There is a risk that the simulated data may
indicators are likely used to assess
the safety performance of different comparative analysis offers valuable insights into the not accurately reflect the variability and
road segments. strengths and weaknesses of different approaches, unpredictability inherent in actual crash occurrences.
3. Monte Carlo Simulation: To
aiding in the selection of the most effective method. 3. Assumption of Methodological Superiority: The
evaluate the accuracy and
efficiency of different procedures 3. Identification of True Positives: The research study asserts that indicators based on empirical
for identifying black spots, the highlights the importance of using indicators based Bayesian estimation PSI represent the best practice
study employs Monte Carlo
simulation. This simulation on empirical Bayesian estimation PSI for identifying for identifying true positives. However, this assertion
technique generates theoretical true positives—sites with genuine safety issues. This may overlook potential advancements or alternative
crash data similar to empirical
data, allowing for the assessment emphasis on accuracy ensures that resources are methodologies that could offer comparable or
of different methods' ability to appropriately allocated to address genuine safety superior performance in specific contexts. A more
 correctly identify hazardous sites. concerns, leading to more effective road safety nuanced evaluation of different approaches could
4. Comparison of Procedures: The interventions. provide a more comprehensive understanding of their
study compares the accuracy and
efficiency of different procedures Key Findings: relative strengths and weaknesses.
based on the evaluation criteria 1. Critical Evaluation of Safety Indicators: The
mentioned earlier. It likely
involves analyzing how well each study critically evaluates various safety indicators
method identifies hazardous sites commonly used for identifying black spots on two-
and whether it produces false
lane rural roads. It emphasizes the need for
positives or false negatives.
5. Recommendation of Best indicators that account for factors such as traffic
Practices: Based on the findings,
volume, segment length, and crash observation
the study provides
recommendations for identifying period to ensure reliable results.
true positives (i.e., genuine black 2. Effectiveness of Empirical Bayesian Estimation:
spots). It suggests that indicators
based on EB estimation PSI are The research underscores the effectiveness of
generally the best practice for empirical Bayesian estimation PSI as the preferred
identifying black spots on two-lane
rural roads. approach for identifying true positive black spots.
This finding highlights the importance of using
robust statistical methods in black spot identification
to minimize the risk of false positives and ensure
targeted safety interventions.

5.

II. ANOTHER VARIABLE (independent or dependent)