>>>dataflair_index =pd.date_range('1\/1\/2000', periods=8)<\/pre>\n- \n
- Series<\/strong><\/li>\n<\/ul>\n
>>>dataflair_s1 = pd.Series(np.random.randn(5), index=['a', 'b', 'c', 'd', 'e'])<\/pre>\n
- \n
- DataFrame<\/strong><\/li>\n<\/ul>\n
>>>dataflair_df1 = pd.DataFrame(np.random.randn(8, 3), index=dataflair_index,columns=['A', 'B', 'C'])<\/pre>\n
- \n
- Panel<\/strong><\/li>\n<\/ul>\n
>>> dataflair_wp1 = pd.Panel(np.random.randn(2, 5, 4), items=['Item1', 'Item2'],major_axis=pd.date_range('1\/1\/2000', periods=5),minor_axis=['A', 'B', 'C', 'D'])<\/pre>\nOutput-<\/strong><\/p>\n
![\"Importing](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/Import-Library-in-Pandas.png\")
<\/a><\/p>\nBefore we dive into Pandas Basic functionalities, let’s discover the File Hierarchy in Pandas<\/a><\/strong><\/p>\n
Now we can start with the basic functionalities of Pandas.<\/p>\n
- \n
- head() function<\/li>\n
- tail() function<\/li>\n
- Attributes<\/li>\n
- Flexible Binary Operations<\/li>\n<\/ol>\n
To view the starting or the ending of a lengthy series, we can use the head() or tail() function.<\/p>\n
1. head() function<\/h3>\n
Let us create a series with 1000 random values<\/p>\n
>>> dataflair = pd.Series(np.random.randn(1000))<\/pre>\n
Using head() function-<\/strong><\/p>\n
>>> dataflair.head()<\/pre>\n
Output-<\/strong><\/p>\n
![\"Pandas](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/head-function-in-Pandas.png\")
<\/a><\/p>\n2. tail() function<\/h3>\n
Now, we use the tail function and set the number of elements to 3:<\/p>\n
>>> dataflair.tail(3)<\/pre>\n
Output-<\/strong><\/p>\n
![\"Applying](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/Tail-Function-in-Pandas.png\")
<\/a><\/p>\nWhat makes Python Pandas unique from other libraries?<\/a><\/strong><\/p>\n
3. Attributes<\/h3>\n
Attributes play a major role in the basic functionality of pandas which helps data scientist for fast analyzing, cleaning, and preparation of data. Pandas objects possess a number of attributes which enable you to access the metadata.<\/p>\n
Shape:<\/strong> It gives the axis dimensions<\/p>\n
Axis labels:\u00a0<\/strong><\/p>\n
- \n
- Series<\/strong>: index (only one axis)<\/li>\n
- DataFrame<\/strong>: index (rows) and columns<\/li>\n
- Panel<\/strong>: major axis, minor axis and items<\/li>\n<\/ol>\n
- \n
- You can safely assign these attributes.<\/strong><\/li>\n<\/ul>\n
>>> dataflair_df1[:2]<\/pre>\n
Output-<\/strong><\/p>\n
![\"attribute](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/Pandas-attribute-example.png\")
<\/a><\/p>\n- \n
- This prints the last two values for the DataFrame<\/strong><\/li>\n<\/ul>\n
>>> dataflair_df1.columns = [x.lower() for x in dataflair_df1.columns]\r\n>>> dataflair_df1<\/pre>\n
Output-<\/strong><\/p>\n
![\"Pandas](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/Print-last-values-of-dataframes-in-pandas.png\")
<\/a><\/p>\n- \n
- Using this function we change the uppercase column names to lowercase.<\/strong><\/li>\n<\/ul>\n
If you have to get the actual data which is inside a Pandas data structure, you only need to use the values property.<\/p>\n
>>> dataflair_s1.values<\/pre>\n
Output-<\/strong><\/p>\n
![\"Changing](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/How-to-Change-upper-column-name-in-Pandas.png\")
<\/a><\/p>\nInput-<\/strong><\/p>\n
>>> dataflair_df1.values<\/pre>\n
Output-<\/strong><\/p>\n
![\"Upper](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/Upper-to-lower-column-name-in-pandas.png\")
<\/a><\/p>\n<\/p>\n
>>> dataflair_wp1.values<\/pre>\n
Output-<\/strong><\/p>\n
![\"Pandas](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/pandas-wp-values.png\")
<\/a><\/p>\n4. Flexible Binary Operations<\/h3>\n
In the binary operations between data structures in pandas<\/strong>, there are two vital points of interest:<\/p>\n
- \n
- Broadcasting behavior between lower dimensional objects and higher dimensional objects<\/li>\n
- Missing data while computing<\/li>\n<\/ul>\n
We will learn how to manage these two issues independently. They can be dealt with simultaneously though.<\/p>\n
4.1 Broadcasting Behaviour<\/h4>\n
For broadcasting behavior, the Series<\/a> input is primary. You can match the index or columns by using the axis() keyword.<\/strong><\/p>\n
>>> dataflair_df = pd.DataFrame({'one' : pd.Series(np.random.randn(3), index=['a', 'b', 'c']),'two' : pd.Series(np.random.randn(4), index=['a', 'b', 'c', 'd']),'three' : pd.Series(np.random.randn(3), index=['b', 'c', 'd'])})\r\n>>> dataflair_df<\/pre>\nOutput-<\/strong><\/p>\n
![\"Using](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/Pandas-Axis-Keywords.png\")
<\/a><\/p>\nInput-<\/strong><\/p>\n
>>> row = dataflair_df.iloc[1]\r\n>>> column = dataflair_df['two']\r\n>>> dataflair_df.sub(row, axis='columns')<\/pre>\n
Output-<\/strong><\/p>\n
![\"Pandas](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/Pandas-iloc-function.png\")
<\/a><\/p>\nPandas are popular in Data Science but Pandas has different applications<\/a> in other sectors too.<\/strong><\/p>\n
>>> dataflair_df.sub(column, axis='index')<\/pre>\n
Output-<\/strong><\/p>\n
![\"Column](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/Column-wise-indexing-in-pandas.png\")
<\/a><\/p>\nInput-<\/strong><\/p>\n
>>> dataflair_df.sub(column, axis=0)\r\n<\/pre>\n
Output-<\/strong><\/p>\n
![\"How](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/pandas-indexed-on-zero-axis.png\")
<\/a><\/p>\n4.1.1 Multi-indexed DataFrames level<\/h5>\n
Using a series, you can align a multi-indexed DataFrame\u2019s level.<\/p>\n
>>> dataflair_dfmi = dataflair_df.copy()\r\n>>> dataflair_dfmi.index = pd.MultiIndex.from_tuples([(1,'a'),(1,'b'),(1,'c'),(2,'a')],names=['first','second'])\r\n>>> dataflair_dfmi.sub(column, axis=0, level='second')<\/pre>\n
Output-<\/strong><\/p>\n
![\"Pandas](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/Pandas-Multi-indexed-DataFrame.png\")
<\/a><\/p>\nIn a Panel, the matching or broadcasting behavior is a little difficult. Hence, the arithmetic methods are used instead, giving you an option to specify the axis of broadcast.<\/p>\n
>>> major_mean = dataflair_wp1.mean(axis='major')\r\n>>> major_mean<\/pre>\n
Output-<\/strong><\/p>\n
![\"pandas](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/pandas-multi-indexed-DataFrame-with-major-axis.png\")
<\/a><\/p>\n>>> dataflair_wp1.sub(major_mean, axis='major')<\/pre>\n
Output-<\/strong><\/p>\n
![\"multi](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/pandas-major-means-axis.png\")
<\/a><\/p>\nSeries and Index support the divmod() built-in function. It takes the floor division along with the modulo operation at the same time and returns a two-tuple of the same type. It returns it as the left-hand side.<\/p>\n
Do you know the benefits offered by Python Pandas<\/a><\/strong>?<\/p>\n
For series<\/h4>\n
>>> dataflair_s = pd.Series(np.arange(10))\r\n>>> dataflair_s<\/pre>\n
Output-<\/strong><\/p>\n
![\"Example](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/Pandas-divmod-function.png\")
<\/a>
\nInput –\u00a0<\/strong><\/p>\n>>> div, rem = divmod(dataflair_s, 3) #Dividing by 3\r\n>>> div<\/pre>\n
0\u00a0 0<\/p>\n
1\u00a0 0<\/p>\n
2\u00a0 0<\/p>\n
3\u00a0 1<\/p>\n
4\u00a0 1<\/p>\n
5\u00a0 1<\/p>\n
6\u00a0 2<\/p>\n
7\u00a0 2<\/p>\n
8\u00a0 2<\/p>\n
9\u00a0 3<\/p>\n
>>> rem<\/pre>\n
![\"Result](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/Applying-Pandas-Divmod-function.png\")
<\/a><\/p>\nFor index<\/h4>\n
>>> dataflair_idx = pd.Index(np.arange(10))\r\n>>> dataflair_idx<\/pre>\n
![\"Indexing](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/Pandas-index-in-Series-.png\")
<\/a><\/p>\n>>> div, rem = divmod(dataflair_idx, 3)\r\n>>> div<\/pre>\n
Int64Index([0,0,0,1,1,1,2,2,2,3], dtype=’int64′)<\/p>\n
>>> rem<\/pre>\n
![\"rem-in-Pandas](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/rem-in-Pandas-using-divmod.png\")
<\/a><\/p>\nWe can do divmod() elementwise as well.<\/p>\n
div, rem = divmod(dataflair_s, [2, 2, 3, 3, 4, 4, 5, 5, 6, 6]) #First element will be divided by 2, second element by 3, third by 3 and so on<\/p>\n
>>> div, rem = divmod(dataflair_s, [2, 2, 3, 3, 4, 4, 5, 5, 6, 6])\r\n>>> div<\/pre>\n
![\"Example](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/Example-of-Pandas-divmod-function.png\")
<\/a><\/p>\n>>> rem<\/pre>\n
![\"rem](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/rem-result-of-divmod-function-in-Pandas.png\")
<\/a><\/p>\n4.2 Missing Values in pandas<\/h3>\n
In DataFrame and Series, the arithmetic function gives you an option of inputting a fill_value, which basically substitutes a value whenever a value is missing from a location. When you add two DataFrame objects, you can treat NaN as 0. However, if both DataFrames are missing that value the result will be NaN. You can still replace it with some other value by using the fillna function later.<\/p>\n
>>> dataflair_df<\/pre>\n
![\"Find](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/Pandas-Missing-Values-1.png\")
<\/a><\/p>\n>>> dataflair_df2 = pd.DataFrame({'one' : pd.Series(np.random.randn(3), index=['a', 'b', 'c']),'two' : pd.Series(np.random.randn(4), index=['a', 'b', 'c', 'd']),'three' : pd.Series(np.random.randn(3), index=['b', 'c', 'd'])})\r\n>>> dataflair_df2<\/pre>\n![\"Get](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/Missing-Values-in-Pandas-3.png\")
<\/a><\/p>\n>>> dataflair_df + dataflair_df2<\/pre>\n
![\"Example](\"https:\/\/data-flair.training\/blogs\/wp-content\/uploads\/sites\/2\/2019\/05\/Missing-values-in-pandas-2.png\")
<\/a><\/p>\nInput<\/strong><\/p>\n
>>> dataflair_df.add(dataflair_df2,fill_value=0) #does the same thing as \u2018+\u2019 operator<\/pre>\n
- Using this function we change the uppercase column names to lowercase.<\/strong><\/li>\n<\/ul>\n
- This prints the last two values for the DataFrame<\/strong><\/li>\n<\/ul>\n
- DataFrame<\/strong>: index (rows) and columns<\/li>\n
- Series<\/strong>: index (only one axis)<\/li>\n
- Panel<\/strong><\/li>\n<\/ul>\n
- DataFrame<\/strong><\/li>\n<\/ul>\n