notebook.community

Edit and run 





In [1]:



    


%ls



    


















    






2015_white_house.csv*  Basics.ipynb*

















In [2]:



    


import numpy as np
import pandas as pd
%matplotlib inline
import matplotlib.pyplot as plt



    











In [3]:



    


white_house = pd.read_csv("2015_white_house.csv")



    











In [4]:



    


white_house.head(5)



    


















    
Out[4]:










  
    

      
      Name
      Status
      Salary
      Pay Basis
      Position Title
    

  
  
    

      0
      Abdullah, Hasan A.
      Detailee
      105960
      Per Annum
      POLICY ADVISOR
    

    

      1
      Abraham, Sabey M.
      Employee
      55000
      Per Annum
      ENERGY AND ENVIRONMENT DIRECTOR FOR PRESIDENTI...
    

    

      2
      Abraham, Yohannes A.
      Employee
      121200
      Per Annum
      SPECIAL ASSISTANT TO THE PRESIDENT AND CHIEF O...
    

    

      3
      Abramson, Jerry E.
      Employee
      155035
      Per Annum
      DEPUTY ASSISTANT TO THE PRESIDENT AND DIRECTOR...
    

    

      4
      Adler, Caroline E.
      Employee
      114000
      Per Annum
      SPECIAL ASSISTANT TO THE PRESIDENT AND DIRECTO...
    

  




















In [5]:



    


white_house.shape



    


















    
Out[5]:








(474, 5)

















In [6]:



    


white_house.iloc[0]



    


















    
Out[6]:








Name              Abdullah, Hasan A.
Status                      Detailee
Salary                        105960
Pay Basis                  Per Annum
Position Title        POLICY ADVISOR
Name: 0, dtype: object

















In [7]:



    


white_house



    


















    
Out[7]:










  
    

      
      Name
      Status
      Salary
      Pay Basis
      Position Title
    

  
  
    

      0
      Abdullah, Hasan A.
      Detailee
      105960
      Per Annum
      POLICY ADVISOR
    

    

      1
      Abraham, Sabey M.
      Employee
      55000
      Per Annum
      ENERGY AND ENVIRONMENT DIRECTOR FOR PRESIDENTI...
    

    

      2
      Abraham, Yohannes A.
      Employee
      121200
      Per Annum
      SPECIAL ASSISTANT TO THE PRESIDENT AND CHIEF O...
    

    

      3
      Abramson, Jerry E.
      Employee
      155035
      Per Annum
      DEPUTY ASSISTANT TO THE PRESIDENT AND DIRECTOR...
    

    

      4
      Adler, Caroline E.
      Employee
      114000
      Per Annum
      SPECIAL ASSISTANT TO THE PRESIDENT AND DIRECTO...
    

    

      5
      Aiyer, Vikrum D.
      Detailee
      134662
      Per Annum
      SENIOR POLICY ADVISOR
    

    

      6
      Alcantara, Elias
      Employee
      65650
      Per Annum
      ASSOCIATE DIRECTOR OF INTERGOVERNMENTAL AFFAIRS
    

    

      7
      Ali, Mohammed I.
      Employee
      42000
      Per Annum
      STAFF ASSISTANT
    

    

      8
      Allen, Angelica P.
      Employee
      50000
      Per Annum
      SPECIAL ASSISTANT TO THE DIRECTOR OF THE OFFIC...
    

    

      9
      Allen, Elizabeth M.
      Employee
      103000
      Per Annum
      SPECIAL ASSISTANT TO THE PRESIDENT FOR MESSAGE...
    

    

      10
      Allen, Jessica L.
      Employee
      42844
      Per Annum
      PRESS ASSISTANT
    

    

      11
      Allison, Ashley R.
      Employee
      97000
      Per Annum
      DEPUTY DIRECTOR OF PUBLIC ENGAGEMENT
    

    

      12
      Amendolare, Vincent C.
      Employee
      42420
      Per Annum
      ANALYST
    

    

      13
      Amuluru, Uma M.
      Detailee
      116804
      Per Annum
      ASSOCIATE COUNSEL
    

    

      14
      Anderson, Amanda D.
      Employee
      126250
      Per Annum
      SPECIAL ASSISTANT TO THE PRESIDENT AND HOUSE L...
    

    

      15
      Anderson, Charles D.
      Employee
      101000
      Per Annum
      SENIOR ADVISOR FOR THE NATIONAL ECONOMIC COUNCIL
    

    

      16
      Aniskoff, Paulette L.
      Employee
      155035
      Per Annum
      DEPUTY ASSISTANT TO THE PRESIDENT AND DIRECTOR...
    

    

      17
      Ashton, Nathaniel R.
      Employee
      42000
      Per Annum
      STAFF ASSISTANT
    

    

      18
      Austin, Jr., Roy L.
      Employee
      160085
      Per Annum
      DEPUTY ASSISTANT TO THE PRESIDENT FOR THE OFFI...
    

    

      19
      Axios, Ashleigh T.
      Employee
      73225
      Per Annum
      DIGITAL CREATIVE DIRECTOR
    

    

      20
      Babajide, Ayotunde T.
      Detailee
      114480
      Per Annum
      SENIOR POLICY ADVISOR
    

    

      21
      Bae, Yena
      Employee
      44000
      Per Annum
      SENIOR ANALYST AND PROJECT MANAGER
    

    

      22
      Baker, Sarah E.
      Employee
      131805
      Per Annum
      SPECIAL ASSISTANT TO THE PRESIDENT AND ASSOCIA...
    

    

      23
      Bansal, Gaurab
      Employee
      120000
      Per Annum
      DEPUTY ASSISTANT TO THE PRESIDENT AND DEPUTY C...
    

    

      24
      Barnes, Desiree N.
      Employee
      42844
      Per Annum
      PRESS ASSISTANT
    

    

      25
      Bartoloni, Kristen A.
      Employee
      70000
      Per Annum
      DEPUTY DIRECTOR OF RESEARCH
    

    

      26
      Beckford, Kevin F.
      Employee
      42000
      Per Annum
      ANALYST
    

    

      27
      Beliveau, Emmett S.
      Employee
      160085
      Per Annum
      ASSISTANT TO THE PRESIDENT AND DIRECTOR OF THE...
    

    

      28
      Benenati, Frank J.
      Employee
      84840
      Per Annum
      ASSISTANT PRESS SECRETARY
    

    

      29
      Bennett, Tabitha R.
      Employee
      50000
      Per Annum
      PRESS LEAD
    

    

      ...
      ...
      ...
      ...
      ...
      ...
    

    

      444
      Vignarajah, Krishanti
      Employee
      90000
      Per Annum
      DIRECTOR OF INTERNATIONAL AFFAIRS AND SENIOR P...
    

    

      445
      Vilfer, Ryan E.
      Employee
      42420
      Per Annum
      STAFF ASSISTANT
    

    

      446
      Vorhaus, David A.
      Employee
      111100
      Per Annum
      SPECIAL ASSISTANT TO THE PRESIDENT AND ADVISOR...
    

    

      447
      Vrazilek, Lauren S.
      Employee
      55000
      Per Annum
      DEPUTY PRESS SECRETARY FOR THE FIRST LADY
    

    

      448
      Wagstaff, Jesica E.
      Employee
      45905
      Per Annum
      LEGISLATIVE ASSISTANT AND ASSISTANT TO THE SEN...
    

    

      449
      Waheed, Manar
      Employee
      80800
      Per Annum
      DEPUTY POLICY DIRECTOR FOR IMMIGRATION
    

    

      450
      Wainscott, Kip F.
      Detailee
      134662
      Per Annum
      SENIOR DIRECTOR OF CABINET AFFAIRS
    

    

      451
      Waldo, Katherine A.
      Employee
      47470
      Per Annum
      TRIP COORDINATOR
    

    

      452
      Wall, Alexander B.
      Employee
      63125
      Per Annum
      DEPUTY DIRECTOR OF ONLINE ENGAGEMENT
    

    

      453
      Walsh, James D.
      Detailee
      171871
      Per Annum
      ASSOCIATE COUNSEL
    

    

      454
      Walsh, Joan L.
      Employee
      160085
      Per Annum
      DEPUTY ASSISTANT TO THE PRESIDENT AND STAFF SE...
    

    

      455
      Wang, Ya W.
      Employee
      85000
      Per Annum
      SPECIAL ASSISTANT AND ADVISOR TO THE CHIEF OF ...
    

    

      456
      Whisenant, Addie M.
      Employee
      78780
      Per Annum
      DIRECTOR OF AFRICAN-AMERICAN MEDIA
    

    

      457
      Wild, Clayton S.
      Employee
      44000
      Per Annum
      PRESIDENTIAL WRITER
    

    

      458
      Wilkinson, Jr., David E.
      Employee
      100000
      Per Annum
      DIRECTOR, OFFICE OF SOCIAL INNOVATION AND CIVI...
    

    

      459
      Williams, Sherman A.
      Employee
      80716
      Per Annum
      ASSISTANT TO THE EXECUTIVE CLERK
    

    

      460
      Winter, Melissa E.
      Employee
      133320
      Per Annum
      DEPUTY ASSISTANT TO THE PRESIDENT AND SENIOR A...
    

    

      461
      Wong, Jacqueline
      Employee
      58075
      Per Annum
      SENIOR POLICY ANALYST
    

    

      462
      Wright, Frank B.
      Employee
      65000
      Per Annum
      DEPUTY DIRECTOR FOR FINANCE
    

    

      463
      Wu, Alexander P.
      Employee
      52520
      Per Annum
      ASSOCIATE DIRECTOR FOR THE MANAGEMENT AND ADMI...
    

    

      464
      Xharda, Klevis
      Employee
      42420
      Per Annum
      EXECUTIVE ASSISTANT
    

    

      465
      Yaros, Stephen G.
      Employee
      42000
      Per Annum
      STAFF ASSISTANT
    

    

      466
      Young, Caitlin E.
      Employee
      63630
      Per Annum
      STENOGRAPHER
    

    

      467
      Young, Jr., Reginald D.
      Employee
      69597
      Per Annum
      SENIOR MANAGEMENT ANALYST
    

    

      468
      Young, Kimberly E.
      Employee
      45000
      Per Annum
      LEGISLATIVE ASSISTANT
    

    

      469
      Young, Stephanie L.
      Employee
      87870
      Per Annum
      ASSOCIATE COMMUNICATIONS DIRECTOR
    

    

      470
      Young, Valerie A.
      Employee
      52667
      Per Annum
      RECORDS MANAGEMENT ANALYST
    

    

      471
      Yudelson, Alex R.
      Employee
      42000
      Per Annum
      STAFF ASSISTANT
    

    

      472
      Zaid, Zaid A.
      Detailee
      158700
      Per Annum
      ASSOCIATE COUNSEL
    

    

      473
      Zients, Jeffrey D.
      Employee
      173922
      Per Annum
      ASSISTANT TO THE PRESIDENT FOR ECONOMIC POLICY...
    

  



474 rows × 5 columns



















In [8]:



    


plt.hist(white_house["Salary"])
plt.show()

Untill now, we overview the functionality that Jupyter notebook provides for us.

How does length of employee titles correlate to salary?





In [9]:



    


position_title = white_house["Position Title"]
title_length = position_title.apply(len)
salary = white_house["Salary"]



    











In [10]:



    


from scipy.stats.stats import pearsonr



    











In [11]:



    


pearsonr(title_length, salary)



    


















    
Out[11]:








(0.53283725149987593, 4.0497353072519681e-36)

















In [12]:



    


plt.scatter(title_length, salary)
plt.xlabel("title length")
plt.ylabel("salary")
plt.title("Title length - Salary Scatter Plot")
plt.show()

How much does the White House pay in total salary?





In [13]:



    


white_house["Salary"].sum()



    


















    
Out[13]:








40225595

Who are the highest and lowest paid staffers?





In [14]:



    


max_salary = white_house["Salary"].max()

max_salary_column = white_house["Salary"] == max_salary

white_house.loc[max_salary_column].reset_index(drop = True)



    


















    
Out[14]:










  
    

      
      Name
      Status
      Salary
      Pay Basis
      Position Title
    

  
  
    

      0
      Breckenridge, Anita J.
      Employee
      173922
      Per Annum
      ASSISTANT TO THE PRESIDENT AND DEPUTY CHIEF OF...
    

    

      1
      Canegallo, Kristie A.
      Employee
      173922
      Per Annum
      ASSISTANT TO THE PRESIDENT AND DEPUTY CHIEF OF...
    

    

      2
      Cushman, Chase M.
      Employee
      173922
      Per Annum
      ASSISTANT TO THE PRESIDENT AND DIRECTOR OF SCH...
    

    

      3
      Earnest, Joshua R.
      Employee
      173922
      Per Annum
      ASSISTANT TO THE PRESIDENT AND PRESS SECRETARY
    

    

      4
      Eggleston, Warren N.
      Employee
      173922
      Per Annum
      ASSISTANT TO THE PRESIDENT AND COUNSEL TO THE ...
    

    

      5
      Fallon, Katherine B.
      Employee
      173922
      Per Annum
      ASSISTANT TO THE PRESIDENT AND DIRECTOR OF THE...
    

    

      6
      Jarrett, Valerie B.
      Employee
      173922
      Per Annum
      SENIOR ADVISOR AND ASSISTANT TO THE PRESIDENT ...
    

    

      7
      Johnson, Broderick D.
      Employee
      173922
      Per Annum
      ASSISTANT TO THE PRESIDENT AND CABINET SECRETARY
    

    

      8
      Kale, Katy A.
      Employee
      173922
      Per Annum
      ASSISTANT TO THE PRESIDENT FOR MANAGEMENT AND ...
    

    

      9
      Keenan, Cody S.
      Employee
      173922
      Per Annum
      ASSISTANT TO THE PRESIDENT AND DIRECTOR OF SPE...
    

    

      10
      McDonough, Denis R.
      Employee
      173922
      Per Annum
      ASSISTANT TO THE PRESIDENT AND CHIEF OF STAFF
    

    

      11
      Monaco, Lisa O.
      Employee
      173922
      Per Annum
      ASSISTANT TO THE PRESIDENT FOR HOMELAND SECURI...
    

    

      12
      Muñoz, Cecilia
      Employee
      173922
      Per Annum
      ASSISTANT TO THE PRESIDENT AND DIRECTOR OF THE...
    

    

      13
      Rhodes, Benjamin J.
      Employee
      173922
      Per Annum
      ASSISTANT TO THE PRESIDENT AND DEPUTY NATIONAL...
    

    

      14
      Rice, Susan E.
      Employee
      173922
      Per Annum
      ASSISTANT TO THE PRESIDENT AND NATIONAL SECURI...
    

    

      15
      Simas, David M.
      Employee
      173922
      Per Annum
      ASSISTANT TO THE PRESIDENT AND DIRECTOR OF THE...
    

    

      16
      Tchen, Christina M.
      Employee
      173922
      Per Annum
      ASSISTANT TO THE PRESIDENT AND CHIEF OF STAFF ...
    

    

      17
      Zients, Jeffrey D.
      Employee
      173922
      Per Annum
      ASSISTANT TO THE PRESIDENT FOR ECONOMIC POLICY...
    

  




















In [15]:



    


min_salary = white_house["Salary"].min()

min_salary_column = white_house["Salary"] == min_salary

white_house.loc[min_salary_column].reset_index(drop = True)



    


















    
Out[15]:










  
    

      
      Name
      Status
      Salary
      Pay Basis
      Position Title
    

  
  
    

      0
      Leary, Kimberlyn R.
      Employee
      0
      Per Annum
      ADVISOR TO THE COUNCIL ON WOMEN AND GIRLS
    

    

      1
      Park, Todd Y.
      Employee
      0
      Per Annum
      ADVISOR FOR TECHNOLOGY

What words are the most common in titles?





In [16]:



    


words = {}
for title in position_title:
    title_list = title.split()
    for word in title_list:
        if word not in words:
            words[word] = 1
        else:
            words[word] += 1



    











In [17]:



    


import operator
sorted_words = sorted(words.items(), key=operator.itemgetter(1), reverse = True)



    











In [18]:



    


sorted_words



    


















    
Out[18]:








[('THE', 218),
 ('ASSISTANT', 215),
 ('TO', 172),
 ('AND', 168),
 ('OF', 155),
 ('DIRECTOR', 146),
 ('PRESIDENT', 135),
 ('SPECIAL', 102),
 ('FOR', 97),
 ('DEPUTY', 86),
 ('SENIOR', 71),
 ('ASSOCIATE', 67),
 ('POLICY', 61),
 ('ADVISOR', 58),
 ('STAFF', 49),
 ('COUNSEL', 37),
 ('PRESIDENTIAL', 37),
 ('AFFAIRS', 35),
 ('ANALYST', 35),
 ('OFFICE', 28),
 ('LEGISLATIVE', 26),
 ('PERSONNEL', 25),
 ('CHIEF', 24),
 ('ENGAGEMENT', 23),
 ('MANAGEMENT', 23),
 ('SECRETARY', 19),
 ('COUNCIL', 18),
 ('PUBLIC', 18),
 ('PRESS', 17),
 ('INFORMATION', 16),
 ('ADVANCE', 16),
 ('ECONOMIC', 16),
 ('NATIONAL', 16),
 ('HOUSE', 15),
 ('SERVICES', 14),
 ('RECORDS', 14),
 ('EXECUTIVE', 14),
 ('OPERATOR', 13),
 ('LIAISON', 12),
 ('COMMUNICATIONS', 12),
 ('INTERGOVERNMENTAL', 11),
 ('CORRESPONDENCE', 10),
 ('LADY', 10),
 ('FIRST', 10),
 ('WHITE', 10),
 ('LEAD', 9),
 ('SCHEDULING', 9),
 ('MANAGER', 8),
 ('MEDIA', 8),
 ('CABINET', 8),
 ('SECURITY', 7),
 ('DIRECTOR,', 7),
 ('DIGITAL', 7),
 ('STRATEGY', 7),
 ('DOMESTIC', 7),
 ('OPERATIONS', 7),
 ('SENATE', 7),
 ('SPEECHWRITER', 7),
 ('COORDINATOR', 6),
 ('ADMINISTRATION', 6),
 ('SUPERVISOR', 6),
 ('OUTREACH', 6),
 ('CLERK', 6),
 ('WRITER', 6),
 ('TRAVEL', 5),
 ('RESEARCH', 5),
 ('PLANNING', 5),
 ('PROJECT', 5),
 ('PRINCIPAL', 5),
 ('ONLINE', 4),
 ('TECHNOLOGY', 4),
 ('STRATEGIC', 4),
 ('ON', 4),
 ('CONTENT', 4),
 ('WOMEN', 4),
 ('TRIP', 4),
 ('AIDE', 4),
 ('POLITICAL', 4),
 ('MOVE!', 3),
 ('SPECIALIST', 3),
 ('GIRLS', 3),
 ("LET'S", 3),
 ('INTERNATIONAL', 3),
 ('ECONOMICS', 3),
 ('CALLIGRAPHER', 3),
 ('REGIONAL', 3),
 ('STENOGRAPHER', 3),
 ('SOCIAL', 3),
 ('STENOGRAPHY', 3),
 ('ETHICS', 3),
 ('SUPPORT', 3),
 ('URBAN', 2),
 ('CONFIRMATIONS', 2),
 ('LEGAL', 2),
 ('FINANCE', 2),
 ('PROGRAM', 2),
 ('JOINING', 2),
 ('AIDS', 2),
 ('TRACKING', 2),
 ('IMMIGRATION', 2),
 ('UNIT', 2),
 ('MOBILITY', 2),
 ('INNOVATION', 2),
 ('FILE', 2),
 ('CLASSIFICATION', 2),
 ('ENERGY', 2),
 ('PROJECTS', 2),
 ('PERSONAL', 2),
 ('SCHEDULER', 2),
 ('SUPERVISOR,', 2),
 ('BROADCAST', 2),
 ('SYSTEMS', 2),
 ('FORCES', 2),
 ('LABOR', 2),
 ('SEARCH', 2),
 ('SPEECHWRITING', 2),
 ('PRESIDENT,', 2),
 ('DOCUMENT', 2),
 ('CLEARANCE', 2),
 ('MESSAGE', 2),
 ('RESEARCHER', 2),
 ('EDUCATION', 2),
 ('VETTER', 2),
 ('REVIEW', 1),
 ('FAMILIES', 1),
 ('KEEPER', 1),
 ('PARTNERSHIPS', 1),
 ('HISPANIC', 1),
 ('VETTING', 1),
 ('WEST', 1),
 ('BROTHER\x92S', 1),
 ('PLANNER', 1),
 ('ACCOMMODATIONS', 1),
 ('RELATIONS', 1),
 ('MONITOR', 1),
 ('MESSAGES', 1),
 ('EVENTS', 1),
 ('VISITORS', 1),
 ('GREETINGS', 1),
 ('OFFICER', 1),
 ('INITIATIVE', 1),
 ('AFRICAN-AMERICAN', 1),
 ('RESPONSE', 1),
 ('RECEPTIONIST', 1),
 ('NUTRITION', 1),
 ('OPPORTUNITY', 1),
 ('LEGISLATION', 1),
 ('SECTOR', 1),
 ('CONSTITUENT', 1),
 ('SPECIALTY', 1),
 ('FAITH-BASED', 1),
 ('BUSINESS', 1),
 ('LINE', 1),
 ('MY', 1),
 ('INTERNSHIP', 1),
 ('CIVIC', 1),
 ('ADVISOR,', 1),
 ('ACTIONS', 1),
 ('MEMBER', 1),
 ('COMMENT', 1),
 ('SWITCHBOARD', 1),
 ('COUNTERTERRORISM', 1),
 ('CREATIVE', 1),
 ('HOMELAND', 1),
 ('OVAL', 1),
 ('NEIGHBORHOOD', 1),
 ('CHANGE', 1),
 ('WING', 1),
 ('HEALTH', 1),
 ('POLICY,', 1),
 ('VIDEO', 1),
 ('AFFAIRS,', 1),
 ('COUNSELOR', 1),
 ('TAX', 1),
 ('DESIGNER', 1),
 ('FRONT', 1),
 ('WORKING', 1),
 ('IMPLEMENTATION', 1),
 ('PARTICIPATION', 1),
 ('CLIMATE', 1),
 ('GIRLS,', 1),
 ('MILITARY', 1),
 ('PRIVATE', 1),
 ('ENVIRONMENT', 1),
 ('PROGRESSIVE', 1),
 ('RECRUITMENT', 1),
 ('JUSTICE,', 1),
 ('WORKFORCE', 1),
 ('FISCAL', 1),
 ('COMPUTER', 1)]

















In [ ]:

Content source: frankbearzou/Data-analysis

Similar notebooks:

notebook.community | gallery | about