Next Wave Mobility

DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING
Phase - 1 Project Presentation on :
NEXT WAVE MOBILITY

Modernizing City Bus Services with
Data Driven Insights
PROJECT ASSOCIATES GUIDE: AMITH SHEKHAR

Sameehana Manvi 1BG20CS086 Associate Professor
Sangamesh R 1BG20CS089 Dept of CSE
Vineeth Phadke 1BG20CS125 BNMIT
INTRODUCTION
Urbanization is exponentially growing in metropolitan cities, and the existing
urban centers lack the necessary resources to sustain this expansion.
Infrastructure , Transportation.
PROBLEM STATEMENT
We seek to revolutionize Bengaluru's urban bus management by introducing data-driven analysis,
dynamic scheduling, and advanced technologies. The goal is to optimize resource utilization,
enhance operational efficiency, and elevate the passenger experience in response to challenges like
inconsistent loads, inefficient scheduling, and a lack of real-time information.
OBJECTIVES
1. Data Collection from ETM APIs:
Establish a robust data retrieval pipeline from ETMs via APIs, ensuring continuous collection for 6 to
8 months to build a diverse dataset.
2.Model Training:
Develop and train a deep learning model for crowd density prediction, optimizing its performance
through the selection of suitable features, architecture, and hyperparameters.
3. Efficient Scheduling:
Implement intelligent scheduling using the trained model to optimize resource efficiency,
minimizing overlaps, and addressing service gaps in the scheduling process.
4.Identify Outliers and Discard:
Implement outlier detection mechanisms and develop strategies for handling outliers to enhance
the accuracy of the model.
BRIDGING THE GAP
Advantages of our solution over other proposed solutions
1.Building on the existing system, where the count of passengers is maintained

by a single person i.e. conductor (unlike the other proposed solution).
2.Eliminating the need for extra training and resources.
3.No additional tasks for passengers and conductors. Ensuring seamless

adaptability.
HARDWARE REQUIREMENTS SOFTWARE REQUIREMENTS
Database Management System

(DBMS):
Backend Development:
Web Development Framework:
Electronic Ticket
vending machine
Machine Learning Framework:
METHODOLOGY
1. Data Collection from ETM APIs:
- Establish API connections for data retrieval.
- Implement a reliable data storage system to
store collected information.
2. Data Collection Duration (6 to 8 months):
- Collect data consistently over the specified
period.
- Ensure data quality and completeness for
meaningful analysis.
METHODOLOGY
3. Model Training:
- Preprocess and clean the collected data.
- Choose and implement appropriate deep
learning and ensemble models.
- Train the models using the prepared dataset.
4. Efficient Scheduling:
- Utilize model predictions to generate efficient
schedules.
- Consider factors like peak hours, routes, and
historical demand.
METHODOLOGY
4.Identify and Discard Outliers:
Implement outlier detection mechanisms to
identify irregularities in the data, such as
erroneous boarding records or anomalies in
passenger counts.
5.Model Validation:
Evaluate the performance of the trained
models on validation datasets.
6.Real-Time Crowd Reflection:
Create a user interface that provides real-
time information on bus crowding.
IMPLEMENTATION
LSTM for Timestamp Selection:
Preprocess data and train an LSTM model.
Divide the dataset into parts.
Quartiles and quantiles within each part to find the best timestamps.
Select the best 30 timestamps from each part.
Monte Carlo Simulation for Best Time:
Apply Monte Carlo simulation to consider all bus stops.

Determine the best time based on simulation results.
AutoEncoders for Outlier Detection:

Implement AutoEncoders to identify outliers in the data.
RESULTS
USE CASE DIAGRAM
CONCLUSION
1. Model Comparison for Bus Arrival Time Estimation 3. Real-World Deployment and Feedback Mechanisms :
Utilization of LSTM, Random Forest, and Implementation of models in real-world scenarios.
Autoencoder models. Utilization of feedback mechanisms for continuous
Analysis of model performance based on improvement.
boarding occurrence data. 4. Effectiveness of Bus Management System
Reduction of passenger wait times through
2. Superior Performance of LSTM and Random Forest optimized scheduling.
Models Adaptation to changing requirements through
LSTM and Random Forest models demonstrate
dynamic scheduling adjustments.
higher accuracy.
Combining models further enhances prediction
capabilities

Next Wave Mobility

Uploaded by

Copyright:

Available Formats

Next Wave Mobility

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Next Wave Mobility

Uploaded by

Copyright:

Available Formats

DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING

Phase - 1 Project Presentation on :

NEXT WAVE MOBILITY

PROJECT ASSOCIATES GUIDE: AMITH SHEKHAR

1.Building on the existing system, where the count of passengers is maintained

2.Eliminating the need for extra training and resources.

3.No additional tasks for passengers and conductors. Ensuring seamless

Database Management System

Web Development Framework:

Divide the dataset into parts.

Apply Monte Carlo simulation to consider all bus stops.

AutoEncoders for Outlier Detection:

You might also like