Capital-Bikeshare-Project

This is a repository for our university training project.

Here are the main features we offer.

Basics:

create_directory

Create directories needed.

entry-counter

Simply counts data entries in a file. Deprecated soon.

season-fitting

Fitting total entries based on season. You need to have MatLab to use it.

CSV-reader

A simple reader for CSV-format data. In this project, a time praser is implemented to convert common time format(MM/DD/YYYY HH:MM) to computer recognizable format(365 days, 1440 minutes array). Then a summary dataset is created.

The usage is:

  g++ csv-reader.cpp -o csv-reader
  (chmod +x csv-reader)
  ./csv-reader

It will generate all 'Gnuplot' files needed.

Generic Plot

Data visualization module to show the data more friendly. We make it automatically by a bash shell.

  chmod +x plotting.sh
  ./plotting.sh

Weather Crawler

Basic Version:

Only generate data from 2014-2017, from a Chinese weather provider.
Data format are not very well.

Pro Version:

Well formated data by Wunderground.
Richer dataset, boarder classfication.

It will generate PNG files.

Map Parser

Creating a mapping between dock no. and geographical coordinates.

The usage is:

  python3 MapParser.py

Weather Regression Tool

Stepwise regression yearly:

Doing stepwise regression from 2011-2016's weather data. Possible data modifications are:

Linear process
Quadartic process
Cubic process
Exponential process
Logarithm process
Fractional process

Full regression:

Select nesserary data in stepwise process. Combine all the data together and verify the model using 2017's data.

Traffic Flux Visualization

This tool contains two parts with different programming languages(C++ / Matlab).

data_slicer

To seperate exactly ONE day in the dataset. No need to call it.

flux_visualization

Firstly, the code will run system calls to data_slicer:

  g++ ./data_slicer.cpp -o ./data_slicer
  ./data_slicer [year] [month] [day]

Then, it will read the need files and generate a real-time flux graph.

Finally, it will generate a statistics matrix showing the place with more than 10 bikes on ride per minute.

Efficiency

About 1.8s to process about 3000000 data entries on a Linux distro.
About 13s to process about 20000000 data entries on a Linux distro.

Notices:

Linux is needed.(untested on Windows or WSL)
Open source data can be fetched on Capital Bikeshare's website.
Python requirements: BeautifulSoup, requests, (time, csv, json).
C++ requirements: g++ version more than 6.0 is recommended.
MatLab requirements: Latest version is recommended. "+" overrided for string concatenation is needed.
Gnuplot requirements: version more than 5.0 is needed.

t0nyx1ang / capital-bikeshare-project Goto Github PK