2022-05-15 22:28 S1 _solved
In [1]:
import pandas as pd
import numpy as np
Pandas -
What is a Dataframe?
DataFrame is a data type provided by the library pandas
In python is the most relevant data type to work with tables and data
List are also important to work with data , as obviously you already know, but we are
going to focus in df (dataframes)
Imagine dataframe as a table created by rows and colummns:
Each row and column is an object type pandas.Series
An object type pandas.Series is a vector (list). In each element contains a label
Create a DataFrame
The main ways to do it:
Using data manually
Lists of lists
Nested dictionaries
Reading the information from .csv file
Using the function pd.read_csv() path of the file is mandatory.
In [1]:
data_lst = [
['A3', 0, -1, 0, 'si'],
['B1', 1, None, 0, 'no'],
['B3', 4, None, 0, 'no'],
['B3', 5, 1, 0, 'si'],
['A1', 4, 0, None, None],
['A3', 1, 2, 1, 'si'],
['C2', 4, 1, 1, 'no']
]
data_lst
[['A3', 0, -1, 0, 'si'],
Out[1]:
['B1', 1, None, 0, 'no'],
['B3', 4, None, 0, 'no'],
['B3', 5, 1, 0, 'si'],
['A1', 4, 0, None, None],
['A3', 1, 2, 1, 'si'],
['C2', 4, 1, 1, 'no']]
In [12]:
col0 = []
for row in data_lst:
col0.append(row[0])
col0
['A3', 'B1', 'B3', 'B3', 'A1', 'A3', 'C2']
Out[12]:
file:///Users/bestricemossberg/Downloads/S1 _solved.html 1/14
,2022-05-15 22:28 S1 _solved
Test
In [7]:
test_df = pd.DataFrame(
data_lst
)
test_df
Out[7]: 0 1 2 3 4
0 A3 0 -1.0 0.0 si
1 B1 1 NaN 0.0 no
2 B3 4 NaN 0.0 no
3 B3 5 1.0 0.0 si
4 A1 4 0.0 NaN None
5 A3 1 2.0 1.0 si
6 C2 4 1.0 1.0 no
In [10]:
test_df = pd.DataFrame(
data_lst,
columns=['A', 'B', 'C', 'D', 'E'],
index=[f'row{i}' for i in range(1, 8)]
)
test_df
Out[10]: A B C D E
row1 A3 0 -1.0 0.0 si
row2 B1 1 NaN 0.0 no
row3 B3 4 NaN 0.0 no
row4 B3 5 1.0 0.0 si
row5 A1 4 0.0 NaN None
row6 A3 1 2.0 1.0 si
row7 C2 4 1.0 1.0 no
DataFrame structure
In [9]:
# .index como .columns are iterable objects
print('ROWS:')
for index in test_df.index:
print(index)
print()
print('COLUMNS:')
for col in test_df.columns:
print(col)
ROWS:
row1
file:///Users/bestricemossberg/Downloads/S1 _solved.html 2/14
,2022-05-15 22:28 S1 _solved
row2
row3
row4
row5
row6
row7
COLUMNS:
A
B
C
D
E
DataFrames can be understood as a matrix of values with an index for rows and an indes for
columns.
Any bi-dimensional subset will be consider as a DataFrame and any one-dimensional will be
consider as a Series data type
Although the DataFrame has explicit indices (labels) for rows and columns, both DataFrame
and Series still have a positional ("hidden") index.
file:///Users/bestricemossberg/Downloads/S1 _solved.html 3/14
, 2022-05-15 22:28 S1 _solved
In [29]:
# first 3 lines
test_df.head(3)
Out[29]: A B C D E
0 A3 0 -1.0 0.0 si
1 B1 1 NaN 0.0 no
2 B3 4 NaN 0.0 no
In [30]:
# last 2 lines
test_df.tail(2)
Out[30]: A B C D E
5 A3 1 2.0 1.0 si
6 C2 4 1.0 1.0 no
Select values
.iloc vs .loc
One of the advantages of the DataFrame is that it allows us to access the elements (rows,
columns, cells...) in two ways:
1. through the position (numerical index), e.g. first row, eighth column, etc...
2. through the labels, e.g. the column named "name", the file with index "FHX129M", etc...
In order to be clear with which type of index we want to use, there are two methods: .iloc
and .loc .
1. .iloc is used to access elements via (numeric) positions.
file:///Users/bestricemossberg/Downloads/S1 _solved.html 4/14
