Analysis of Kaohsiung Bike Rental System

Raw   Data

Ability of Big Data-analyzing & UI Design

We   hope   to

• Find statistical correlation
  between demands and multiple variables

Then   we   can

• Improve efficiency of bike allocation

• Suggest new station locations

Define   “The   Demands”

• Counts:
  amount of usage given 30 minutes

• Duration:
  sum of total usage time given 30 minutes

Methodology

• Functional Data Analysis(FDA):
  
  \(Demands=f(t), t=[0, 24]\)

• Discrete-to-continuous conversion

• Better and more appropriate to fit “Big Data”
  Dimension reduction \(=>\) Reduce computation cost

Indivisual   case   I(Demand Gap?)

Indivisual   case   I(More new stations)

Indivisual   case   II(Inefficiency?)

Indivisual   case   II
(Reallocation between both stations)

Indivisual   case   III(Weekend Effect?)

Indivisual   case   III(More new stations)

Thanks.