Large Scale Transportation Data Analysis and Distributed Computational Pipeline for Optimal Metro Passenger Flow Prediction

Title

Large Scale Transportation Data Analysis and Distributed Computational Pipeline for Optimal Metro Passenger Flow Prediction

Creator

Sajanraj, T D

Contributor

S, Raghavendra and Mulerikkal, Jaison Paul

Description

Transportation has a signifcant impact on controlling traffc around a busy city. Among the transport system, metro rails became the backbone by operating above the traffc. For this reason, we have to take special consideration of the passenger and#64258;ow in the transport system and, by understanding the needs, take timely actions for smooth running. Every metro system stores information about the and#64258;ow of passengers in the form of transactions known as Automatic Fare Collection (AFC) data. For this research, AFC data is taken as the primary newlinesource of information to identify the passenger and#64258;ow within the metro rail platform. Each metro system generates massive data throughout its running period and stores data within the system and considering the size of data generated, the analytic platform has to process them in a distributed paradigm to handle quotBig Dataquot. Artifcial Intelligence (AI) algorithms can derives information, insights, and patterns from this data. The patterns in time series can be identifed from the passenger and#64258;ow data using exploratory analysis. The step is an essential step in data science for understanding the underlying properties of the raw data. The research uses a data platform with a distributed computing and storage mechanism called the JP-DAP. The research leverages the above mentioned platform to extract passenger and#64258;ow data from AFC Ticketing data. After the data engineering, the results of passenger and#64258;ow information underwent further visualization and trend analysis. Based on the facts or patterns identifed from the passenger and#64258;ow information, a decision is taken for forecasting. The initial study will reveal the characteristics of metro usage and practices within the system and fnally derive a solution with machine learning-based forecasting method. The passenger and#64258;ow newlineforecasts based on the above patterns depend on factors like seasonality, trends, cyclicity, location, events, and random effects.

Source

Author's Submission

Date

2024-01-01

Publisher

Christ(Deemed to be University)

Subject

Computer Science and Engineering

Rights

Open Access

Relation

61000364

Format

PDF

Language

English

Type

PhD

Identifier

http://hdl.handle.net/10603/585429