1. Importing Data

Let’s get the data to our local machines

wget https://raw.githubusercontent.com/lakshya90/DataScience101/master/titanic.csv (curl -O for mac)

Let’s import relevant libraries

import pandas as pd

Let’s read the file into a dataframe

df = pd.read_csv('titanic.csv')

2. Viewing/Inspecting Data

Let’s see number of rows and columns

df.shape       # Output : 891,12

Let’s see top 5 rows

df.head(5)

Let’s see bottom 5 rows

df.tail(5)

Let’s see some more information on the rows and columns

df.info()       # Output : 891,12

Let’s see some statistics for numerical fields

df.describe()

Let’s see unique counts for each columns

df['Sex'].value_counts(); 
df['Survived'].value_counts();
df['Pclass'].value_counts() 

3. Selection

Let’s see the type when we pick a column

d = df['PassengerId']; type(d)   #<class ‘pandas.core.series.Series’>  

Let’s see any one row features (Pick any number between 0 and 890)

df.iloc[654,:]

4. Data Cleaning

A few columns have NaN values. How do I know?

df.describe() #Check count of all features
df_a = df; df_a['Age'] = df_a['Age'].fill na(df_a['Age'].mean())
df_a['Age'].count() #891 from previous 714

Let’s drop some irrelevant columns

df.drop(['PassengerId','Name','Ticket', 'Cabin'], axis=1)

Let’s see how many null values exist for a certain column

pd.isnull(df['Cabin'])
df[df['Embarked'].isnull()]

5. Filter, Sort, Group By

Let’s see the count of upper class values

df['Pclass'].value_counts(); s = df['Pclass'] < 2; s.value_counts()

Let’s sort all ages in ascending order

df.sort_values('Age')

Let’s see Survivors grouped by Sex.

df.groupby('Sex')['Survived'].value_counts()
df.groupby('Sex').Survived.mean()

6. Statistics

Let’s see some statistics for categorical fields

df.describe(include=['O'])

Let’s see some statistics for a few columns

df['Age'].mean()    #Mean of all values of the feature ‘Age’
df['Cabin'].count()    #Count of all valid ‘Cabin’ values
df['Fare'].max()   #Maximum fare paid for the ticket

7. Exporting the Data

Let's write the data to a csv file

df.to_csv('output.csv')