26  Intro to PD

import pandas as pd
import numpy as np
from pandas import Series, DataFrame

26.1 Series

26.1.1 Creation

obj = pd.Series([4, 7, -5, 3])
type(obj)
obj
0    4
1    7
2   -5
3    3
dtype: int64
# Array Representation
obj.array
type(obj.array)

# Index
obj.index
type(obj.index)
pandas.core.indexes.range.RangeIndex

Series with index

obj2 = pd.Series([4, 7, -5, 3], index=["d", "b", "a", "c"])
obj2

# Index
obj2.index
Index(['d', 'b', 'a', 'c'], dtype='object')

From Dict

sdata = {"Ohio": 35000, "Texas": 71000, "Oregon": 16000, "Utah": 5000}
obj3 = pd.Series(sdata)
obj3
Ohio      35000
Texas     71000
Oregon    16000
Utah       5000
dtype: int64

Back to Dict, List, etc.

obj3.to_dict()
obj3.to_list()
[35000, 71000, 16000, 5000]

26.1.2 Subset Series

# Position
obj2[0] 

# Index
obj2["d"]

# list of indices
obj2[["d", "a"]]

# Logical
obj2[obj2 > 0]

# Assign
obj2["c"] = 10
obj2
/var/folders/70/7wmmf6t55cb84bfx9g1c1k1m0000gn/T/ipykernel_19261/748481342.py:2: FutureWarning: Series.__getitem__ treating keys as positions is deprecated. In a future version, integer keys will always be treated as labels (consistent with DataFrame behavior). To access a value by position, use `ser.iloc[pos]`
  obj2[0]
d     4
b     7
a    -5
c    10
dtype: int64

26.1.3 Vectorized Operation

obj2 * 2
d     8
b    14
a   -10
c    20
dtype: int64

Series is as a fixed-length, ordered dictionary

"b" in obj2
True

26.1.4 Missing Value

x = pd.Series([1, 2, None])
x
# Check NA
x.isna()
x.notna()
0     True
1     True
2    False
dtype: bool

26.1.5 Auto-align arithmatic

states = ["California", "Ohio", "Oregon", "Texas"]
obj4 = pd.Series(sdata, index=states)
obj3
obj4
California        NaN
Ohio          35000.0
Oregon        16000.0
Texas         71000.0
dtype: float64

Data alignment features

obj3 + obj4
California         NaN
Ohio           70000.0
Oregon         32000.0
Texas         142000.0
Utah               NaN
dtype: float64

26.1.6 name Attribute

obj4.name = "population"
obj4.index.name = "state"
obj4
state
California        NaN
Ohio          35000.0
Oregon        16000.0
Texas         71000.0
Name: population, dtype: float64

26.2 DataFrame

See Comparison with R data.frame

it can be thought of as a dictionary of Series all sharing the same index.

26.2.1 Creation

data = {"state": ["Ohio", "Ohio", "Ohio", "Nevada", "Nevada", "Nevada"],
        "year": [2000, 2001, 2002, 2001, 2002, 2003],
        "pop": [1.5, 1.7, 3.6, 2.4, 2.9, 3.2]}
        
frame = pd.DataFrame(data)
type(frame)
frame
state year pop
0 Ohio 2000 1.5
1 Ohio 2001 1.7
2 Ohio 2002 3.6
3 Nevada 2001 2.4
4 Nevada 2002 2.9
5 Nevada 2003 3.2

With specified column arrangement

frame2 = pd.DataFrame(data, columns=["year", "state", "pop", "debt"])
frame2

# Column names
frame2.columns.to_list()
['year', 'state', 'pop', 'debt']
# Shape
frame.shape
# Head
frame.head()
# Tail
frame.tail(3)
state year pop
3 Nevada 2001 2.4
4 Nevada 2002 2.9
5 Nevada 2003 3.2

Nested dictionary

populations = {"Ohio": {2000: 1.5, 2001: 1.7, 2002: 3.6},
               "Nevada": {2001: 2.4, 2002: 2.9}}

pd.DataFrame(populations)
Ohio Nevada
2000 1.5 NaN
2001 1.7 2.4
2002 3.6 2.9

26.2.2 Rows & Colums

frame.shape
(6, 3)
frame.columns
Index(['state', 'year', 'pop'], dtype='object')

26.2.3 Subset

frame2.head()
year state pop debt
0 2000 Ohio 1.5 NaN
1 2001 Ohio 1.7 NaN
2 2002 Ohio 3.6 NaN
3 2001 Nevada 2.4 NaN
4 2002 Nevada 2.9 NaN

26.2.4 Select -> [[

R

select(df, col1, col2)

Py

df[['col1', 'col2']]

Select 1 column

# Quote or unquote
frame2["state"]

frame2["state"] is frame2.state

type(frame2.state) # Series
pandas.core.series.Series

Select multiple column

frame2sub = frame2[["state", "pop"]]
type(frame2sub) # DF
frame2sub
state pop
0 Ohio 1.5
1 Ohio 1.7
2 Ohio 3.6
3 Nevada 2.4
4 Nevada 2.9
5 Nevada 3.2

26.2.5 Slice

R

slice(df, 1:10)

Py

df.iloc[:9]

.iloc: integer location

frame.iloc[:3] # First 3 rows
state year pop
0 Ohio 2000 1.5
1 Ohio 2001 1.7
2 Ohio 2002 3.6 
# Rows, Colums
frame.iloc[[0,1], [0, 2]]
state pop
0 Ohio 1.5
1 Ohio 1.7

.loc: Location by Labels

frame.loc[[0, 1], ["state", "year"]]
state year
0 Ohio 2000
1 Ohio 2001

Slicing also works

frame.loc[0:2, "state":"pop"]
state year pop
0 Ohio 2000 1.5
1 Ohio 2001 1.7
2 Ohio 2002 3.6

26.2.6 Filter -> query

R

filter(df, col1 == 1, col2 == 1)

Py

df.query('col1 == 1 & col2 == 1')
frame2

frame2[frame2.state == "Ohio"]
# Or
frame2.query('state == "Ohio"')

frame2.query('pop > 1.5')
year state pop debt
1 2001 Ohio 1.7 NaN
2 2002 Ohio 3.6 NaN
3 2001 Nevada 2.4 NaN
4 2002 Nevada 2.9 NaN
5 2003 Nevada 3.2 NaN

26.2.7 Mutate -> assign

R: mutate(df, c=a-b)

Py: df.assign(c=df['a']-df['b'])

frame2.assign(pop2 = frame2["pop"] * 2)
year state pop debt pop2
0 2000 Ohio 1.5 NaN 3.0
1 2001 Ohio 1.7 NaN 3.4
2 2002 Ohio 3.6 NaN 7.2
3 2001 Nevada 2.4 NaN 4.8
4 2002 Nevada 2.9 NaN 5.8
5 2003 Nevada 3.2 NaN 6.4 
frame2["debt"] = np.arange(6.)
frame2
year state pop debt
0 2000 Ohio 1.5 0.0
1 2001 Ohio 1.7 1.0
2 2002 Ohio 3.6 2.0
3 2001 Nevada 2.4 3.0
4 2002 Nevada 2.9 4.0
5 2003 Nevada 3.2 5.0

Assign by index of series

val = pd.Series([-1.2, -1.5, -1.7], index=[2, 4, 5])
val
2   -1.2
4   -1.5
5   -1.7
dtype: float64
frame2.debt = val
frame2
year state pop debt
0 2000 Ohio 1.5 NaN
1 2001 Ohio 1.7 NaN
2 2002 Ohio 3.6 -1.2
3 2001 Nevada 2.4 NaN
4 2002 Nevada 2.9 -1.5
5 2003 Nevada 3.2 -1.7 
frame2["pop2"] = frame2["pop"] * 2
frame2
year state pop debt pop2
0 2000 Ohio 1.5 NaN 3.0
1 2001 Ohio 1.7 NaN 3.4
2 2002 Ohio 3.6 -1.2 7.2
3 2001 Nevada 2.4 NaN 4.8
4 2002 Nevada 2.9 -1.5 5.8
5 2003 Nevada 3.2 -1.7 6.4

26.2.8 Rename

rename(df, col_one = col1)
df.rename(columns={'col1': 'col_one'})
frame2.rename(columns = {'year': 'yr'})
yr state pop debt pop2
0 2000 Ohio 1.5 NaN 3.0
1 2001 Ohio 1.7 NaN 3.4
2 2002 Ohio 3.6 -1.2 7.2
3 2001 Nevada 2.4 NaN 4.8
4 2002 Nevada 2.9 -1.5 5.8
5 2003 Nevada 3.2 -1.7 6.4

26.2.9 Summarize

frame2.describe()
year pop debt pop2
count 6.000000 6.000000 3.000000 6.000000
mean 2001.500000 2.550000 -1.466667 5.100000
std 1.048809 0.836062 0.251661 1.672124
min 2000.000000 1.500000 -1.700000 3.000000
25% 2001.000000 1.875000 -1.600000 3.750000
50% 2001.500000 2.650000 -1.500000 5.300000
75% 2002.000000 3.125000 -1.350000 6.250000
max 2003.000000 3.600000 -1.200000 7.200000

26.2.10 Group by, Summarize

frame2.head()
year state pop debt pop2
0 2000 Ohio 1.5 NaN 3.0
1 2001 Ohio 1.7 NaN 3.4
2 2002 Ohio 3.6 -1.2 7.2
3 2001 Nevada 2.4 NaN 4.8
4 2002 Nevada 2.9 -1.5 5.8 
frame2.groupby('state').agg({'pop': 'mean'})
pop
state
Nevada 2.833333
Ohio 2.266667

26.2.11 Count

frame2.value_counts("year")
year
2001    2
2002    2
2000    1
2003    1
Name: count, dtype: int64
frame2.value_counts(["year", "state"], sort=True)
year  state 
2000  Ohio      1
2001  Nevada    1
      Ohio      1
2002  Nevada    1
      Ohio      1
2003  Nevada    1
Name: count, dtype: int64