Route optimization by avoiding unwanted places using unsupervised machine learning

This thesis examines how sophisticated route planning systems employ uncontrolled machine learning algorithms to avoid unpleasant sites users have indicated. The focus is on Density￾Based Spatial Clustering of Applications with Noise (DBSCAN). Traditional route algorithms aim to minimize journey time and distance. These approaches do not always include user preferences, safety concerns, or other subjective or situation-specific variables. This thesis developed and tested a novel route system that combines grouping and location data analytics to locate and avoid locations users do not want to go, such as busy neighbourhoods, casinos, pubs, and other unpleasant venues. DBSCAN can manage noise and locate any-shaped groups, making it ideal for this purpose. Even if undesired sites are unevenly distributed, it can analyze geographical data effectively. After defining density limitations based on user-defined parameters, DBSCAN finds dense clusters of locations based on their geographical proximity. This grouping strategy permits routes to be adjusted depending on data, unlike standard systems that require cutoff areas or zones. The thesis technique analyses the issue statement, revealing the requirement for bespoke route solutions that exceed efficiency criteria. The mathematical origins of DBSCAN and how the algorithm's core assumptions allow it to locate key geographical groupings in enormous volumes of data are also examined. The solution's design and route calculation using regional inputs, user choices, and grouping results are given. The system constantly recalculates the optimum routes to minimise exposure to these undesirable sites while fulfilling travel speed targets. The thesis evaluates system efficacy using experimental testing and scenario-based models. These tests evaluate how effectively the new pathways avoid key groups, how fast and accurately the clustering approach detects them, and how much they vary trip time and distance. The paper highlights its drawbacks, such as calculations being difficult and grouping variables being susceptible to urban contexts. This thesis reveals that personalised and context-aware navigation systems may leverage autonomous machine learning techniques like DBSCAN to determine routes, making them an exciting new subject to examine. The findings demonstrate the need of adding location-type-specific user constraints, which increase directing control and flexibility. Aligning pathways with users' preferred avoidance options makes them happy and allows for the creation of flexible navigation systems that can adjust to social and natural changes.