In [1]:

    

import pandas as pd


    

In [2]:

    

%matplotlib inline


    

In [3]:

    

upcat_data = pd.read_csv("./passers.csv")
upcat_data[:5]


    

    
Out[3]:






  

    

      

      
Name
      
Campus
      
Course
    
  
  

    

      
0
      
        AALA, ANJILA BIANCA LAO
      
    MANILA
      
     BS INDUSTRIAL PHARMACY (5 YRS)
    
    

      
1
      
        AALA, IRMA RUTH PALUBON
      
          
      
                     **Pending Case
    
    

      
2
      
  AALA, JADE GABRIELLE MAGLUNOB
      
   VISAYAS
      
 DEGREE PROGRAM WITH AVAILABLE SLOT
    
    

      
3
      
 ABABA, ETHEL MAY GRACE LACERNA
      
   VISAYAS
      
                      BA PSYCHOLOGY
    
    

      
4
      
  ABACA, HANZ CHRISTIAN CALILAP
      
 LOS BAÑOS
      
                BS COMPUTER SCIENCE
    
  

In [4]:

    

course_data = upcat_data[['Course']]
course_data[:10]


    

    
Out[4]:






  

    

      

      
Course
    
  
  

    

      
0
      
                    BS INDUSTRIAL PHARMACY (5 YRS)
    
    

      
1
      
                                    **Pending Case
    
    

      
2
      
                DEGREE PROGRAM WITH AVAILABLE SLOT
    
    

      
3
      
                                     BA PSYCHOLOGY
    
    

      
4
      
                               BS COMPUTER SCIENCE
    
    

      
5
      
                DEGREE PROGRAM WITH AVAILABLE SLOT
    
    

      
6
      
 BA EUROPEAN LANGUAGESWaitlisted: BS COMPUTER S...
    
    

      
7
      
                      BS DEVELOPMENT COMMUNICATION
    
    

      
8
      
                             BS BUSINESS ECONOMICS
    
    

      
9
      
                        BA BROADCAST COMMUNICATION
    
  

In [5]:

    

course_series = upcat_data['Course'].astype('category')


    

In [6]:

    

course_series.value_counts()[:10]


    

    
Out[6]:





DEGREE PROGRAM WITH AVAILABLE SLOT    3114
**Pending Case                        1538
BS BIOLOGY                             904
BS COMPUTER SCIENCE                    742
BS MATHEMATICS                         341
BS MANAGEMENT                          321
BA POLITICAL SCIENCE                   241
BS CHEMICAL ENGINEERING (5 YRS)        231
BA PSYCHOLOGY                          223
BS CIVIL ENGINEERING (5 YRS)           218
dtype: int64

In [7]:

    

course_series.value_counts()[2:10].plot(kind='bar')


    

    
Out[7]:





<matplotlib.axes._subplots.AxesSubplot at 0xaf6f320>






    

In [8]:

    

names = upcat_data['Name'].astype('category')
names[:10]


    

    
Out[8]:





0            AALA, ANJILA BIANCA LAO
1            AALA, IRMA RUTH PALUBON
2      AALA, JADE GABRIELLE MAGLUNOB
3     ABABA, ETHEL MAY GRACE LACERNA
4      ABACA, HANZ CHRISTIAN CALILAP
5      ABACAN, AUSTIN SYMON BULACLAC
6     ABACAN, JOHN CHRISTIAN BARITUA
7    ABAD, ANGELICA SARAH ESTANISLAO
8        ABAD, BIANCA DENISE MENDOZA
9       ABAD, DUSTIN PAOLO SALVANERA
Name: Name, dtype: category
Categories (14988, object): [AALA, ANJILA BIANCA LAO < AALA, IRMA RUTH PALUBON < AALA, JADE GABRIELLE MAGLUNOB < ABABA, ETHEL MAY GRACE LACERNA ... ZULUETA, PIA YSABEL BANAS < ZULUETA, SANDRA MAE CARAAN < ZUNIGA, CINDERELLA COJAS < ZUNIGA, DARYLL KENDRICK CABIGAO]

In [9]:

    

upcat_data.insert(loc=0, column='Familyname', value='')
upcat_data[:5]


    

    
Out[9]:






  

    

      

      
Familyname
      
Name
      
Campus
      
Course
    
  
  

    

      
0
      
 
      
        AALA, ANJILA BIANCA LAO
      
    MANILA
      
     BS INDUSTRIAL PHARMACY (5 YRS)
    
    

      
1
      
 
      
        AALA, IRMA RUTH PALUBON
      
          
      
                     **Pending Case
    
    

      
2
      
 
      
  AALA, JADE GABRIELLE MAGLUNOB
      
   VISAYAS
      
 DEGREE PROGRAM WITH AVAILABLE SLOT
    
    

      
3
      
 
      
 ABABA, ETHEL MAY GRACE LACERNA
      
   VISAYAS
      
                      BA PSYCHOLOGY
    
    

      
4
      
 
      
  ABACA, HANZ CHRISTIAN CALILAP
      
 LOS BAÑOS
      
                BS COMPUTER SCIENCE
    
  

In [10]:

    

def process_names(text):
    familyname, name = text.split(',')
    return pd.Series([familyname, name])


    

In [11]:

    

upcat_data[['Familyname', 'Name']] = upcat_data.Name.apply(process_names)
upcat_data[:5]


    

    
Out[11]:






  

    

      

      
Familyname
      
Name
      
Campus
      
Course
    
  
  

    

      
0
      
  AALA
      
        ANJILA BIANCA LAO
      
    MANILA
      
     BS INDUSTRIAL PHARMACY (5 YRS)
    
    

      
1
      
  AALA
      
        IRMA RUTH PALUBON
      
          
      
                     **Pending Case
    
    

      
2
      
  AALA
      
  JADE GABRIELLE MAGLUNOB
      
   VISAYAS
      
 DEGREE PROGRAM WITH AVAILABLE SLOT
    
    

      
3
      
 ABABA
      
  ETHEL MAY GRACE LACERNA
      
   VISAYAS
      
                      BA PSYCHOLOGY
    
    

      
4
      
 ABACA
      
   HANZ CHRISTIAN CALILAP
      
 LOS BAÑOS
      
                BS COMPUTER SCIENCE
    
  

In [12]:

    

familynames = upcat_data['Familyname'].astype('category')
familynames[:5]


    

    
Out[12]:





0     AALA
1     AALA
2     AALA
3    ABABA
4    ABACA
Name: Familyname, dtype: category
Categories (8465, object): [AALA < ABABA < ABACA < ABACAN ... ZUBELDIA < ZULLA < ZULUETA < ZUNIGA]

In [13]:

    

familynames.value_counts()


    

    
Out[13]:





REYES        95
TAN          94
CRUZ         89
DELA CRUZ    83
SANTOS       80
GARCIA       80
LIM          73
RAMOS        68
MENDOZA      65
FERNANDEZ    54
CHUA         52
DE GUZMAN    51
BAUTISTA     50
GONZALES     49
ONG          46
...
SIGNEY      1
SIGLOS      1
LARRACAS    1
LASQUITE    1
LASALETA    1
LASAM       1
LASARA      1
SIERVO      1
LASCO       1
LASCUNA     1
SIENES      1
LASPINAS    1
LASQUERO    1
LASQUETE    1
JUMSALI     1
Length: 8465, dtype: int64

In [14]:

    

familynames.value_counts()[:10].plot(kind='bar')


    

    
Out[14]:





<matplotlib.axes._subplots.AxesSubplot at 0x10db4390>






    

In [15]:

    

campuses = upcat_data['Campus'].astype('category')
campuses[:10]


    

    
Out[15]:





0       MANILA
1             
2      VISAYAS
3      VISAYAS
4    LOS BAÑOS
5      DILIMAN
6      DILIMAN
7    LOS BAÑOS
8      DILIMAN
9      DILIMAN
Name: Campus, dtype: category
Categories (11, object): [BAGUIO < CEBU < DILIMAN < LOS BAÑOS ... PAMPANGA < TACLOBAN < VISAYAS < ]

In [16]:

    

campuses.value_counts()[:5]


    

    
Out[16]:





DILIMAN      5154
LOS BAÑOS    2530
             1538
VISAYAS      1523
BAGUIO       1281
dtype: int64

In [17]:

    

campuses.value_counts()[:5].plot(kind='bar')


    

    
Out[17]:





<matplotlib.axes._subplots.AxesSubplot at 0x10d1fb38>






    

In [ ]: