Work

“I have found that small wins, small projects, small differences often make huge differences.”

Rosabeth Moss Kanter

IU Bus System Route Optimization

Problem Definition: IU Bus Service has the messy and scattered data for Fall 2014 and Spring 2015 semester with
  1. driver bus schedule
  2. route information
  3. bus stops
  4. actual time for bus to stop
  5. scheduled time for bus to stop
  6. Incomplete Weather Data
  7. driver's information
  8. disstorted passenger's count
    9. Total Routes : A,B,E,X
Goals:
  1. Determining time deviation (actual-schedule)
  2. Determining effects of passenger's count, weather and stop timing
  3. Determining average travel time and average error for any two given stops for Spring and Fall
  4. Optimization to carry maximum passengers, thereby maintaining the minimum frequency service
  5. Maximize On-route time for all the 4 buses
  6. Route optimization
  7. Schedule Optimization
  8. Dynamic scheduling
Advisor: Dr. Mehmet Dalkilic
Project Contributors: Vipul Munot, Manashree Rao and Ashwin Nimhan.
Technologies:MS SQL, MongoDB and Python. Weather Data: http://forecast.io/

Our Recommendation: We propose a 3 tier architecture with capabalities for realtime straeming of gps data and agregatoins on the fly for better understanding of resource utilization. Dashboard for realtime analysis of the current state of system. A light weight mid-tier to cater to the dashboard and the data-storage tier for stream, and bulk data analysis. We also suggest a dynamic routing and scheduling approach to cope up with the morning rush and then the subsequent lull in the afternoon. Also night time schedule needs to be revisited as the utilization falls dramatically which is supported by zero dwell time and significant negative delay observations at stops stated in the report.

Full Report