deepeye_pack

Update - 2018/10/22 v0.0.2

1. read_csv_handld_changedate revise to pandas method
2. both the import methods require Mandatory table_info to specify the table
3. mysql_handle now change to pandas dataframe verison
4. from_mysql also changes to para-method, pass port/user/db... instead of query and MySQLdb conn

Description

1. This is a Python package for DeepEye API,can easily visualize data without too much effort. And provide with really simple usage
2. the DeepEye system: https://github.com/TsinghuaDatabaseGroup/DeepEye/tree/master/APIs_Deepeye

Installation

1. Python 2.7
2. MySQL 5.7
3. Packages
   1. mysqldb binary packages for windows: link1:https://www.lfd.uci.edu/~gohlke/pythonlibs/
      link2:https://sourceforge.net/projects/mysql-python/
      • Download 'MySQL-python' and choose the right version for it
      • Install the .whl by wheel install
      • there is a back up version in this repository under 'mysqldb' folder
   2. numpy(latest version)
   3. pandas(latest version) above ver 0.23.0

Usage

1. Initial
   1. example code:

   import deepeye_pack
   
   #create a deepeye_pack class that wraps everything
   dp = deepeye_pack.deepeye('demo') # the name here doesnt actually matter
   
   # then user needs to input table info
   # as in table_info(table_name,column_names,column_types)
   dp.table_info('electricity',['city','date','electricity(kWh)'],['varchar','date','float'])
   

   2. the column_types that supported by deepeye_pack are specified as below:
      1. numerical: int, float, double
      2. temporal: date, datetime, year
      3. categorical: char, varchar
2. Import
   1. from_mysql()

      # call the from_mysql() function
      dp.from_mysql(host='localhost',port=3306,user='root',passwd='ppww',db='deepeye', query='SELECT * FROM `table_name`')
      

   2. from_csv()

      path = "file.csv" # the path where the file located
      dp.from_csv(path)
      

3. Visualization

   # choose one from three
   dp.learning_to_rank()
   dp.partial_order()
   dp.diversified_ranking()
   

4. Output

   # can use several different methods at the same time
   dp.to_print_out()
   dp.to_single_json()
   dp.to_multiple_jsons()
   dp.to_multiple_htmls()
   dp.to_single_html()