1 Introduction

It will always happen that you have an incorrect or unsuitable data type and you have to change it. It is therefore worthwhile to familiarize yourself with the conversion methods that exist in python.

For this post the dataset Auto-mpg from the statistic platform “Kaggle” was used. You can download it from my GitHub Repository.

2 Loading the libraries and the data

import pandas as pd
import numpy as np

#for chapter 6
from datetime import datetime

cars = pd.read_csv("path/to/file/auto-mpg.csv")

3 Overview of the existing data types

Numeric:

• integer: Positive or negative whole numbers (without a fractional part)
• float: Any real number with a floating point representation in which a fractional component is denoted by a decimal symbol or scientific notation
• complex number: A number with a real and imaginary component represented as x+zj. x and z are floats and j is -1(square root of -1 called an imaginary number)

Boolean

Data with one of two built-in values ‘True’ or ‘False.’

Sequence Type

• string: A string value is a collection of one or more characters put in single, double or triple quotes.
• list: A list object is an ordered collection of one or more data items, not necessarily of the same type, put in square brackets.
• tuple: A Tuple object is an ordered collection of one or more data items, not necessarily of the same type, put in parentheses.

Dictionary

A dictionary object is an unordered collection of data in a key:value pair form. A collection of such pairs is enclosed in curly brackets. For example: {1:“Sven”, 2:“Tom”, 3:“Eva”, 4: “Will”}

You can check data types in python like this:

type(1234)

type(55.50)

type(6+4j)

type("hello")

type([1,2,3,4])

type((1,2,3,4))

type({1:"one", 2:"two", 3:"three"})

4 Type Conversion

First of all let’s have a look at the data types of our dataframe cars:

cars.dtypes

cars['mpg'].dtypes

cars['mpg'] = cars['mpg'].astype('float32')
cars['mpg'].dtypes

cars['mpg'].head()

cars['mpg'] = cars['mpg'].astype('int64')

cars['mpg'].head()

cars["horsepower"] = pd.to_numeric(cars.horsepower, errors='coerce')

cars['horsepower'].dtypes

cars['horsepower'].head()

cars['horsepower'] = cars['horsepower'].astype('Int64')

cars['horsepower'].dtypes

cars['horsepower'].head()

5 Conversion of multiple variables

cars[['cylinders', 'weight']].dtypes

cars[['cylinders', 'weight']] = cars[['cylinders', 'weight']].astype('int32')
cars[['cylinders', 'weight']].dtypes

6 Conversion of date and time variables

df = pd.DataFrame({'year': [2015, 2016],
                   'month': [2, 3],
                   'day': [4, 5],
                   'hour': [9, 11],
                   'minutes': [22, 50],
                   'seconds': [12, 8]})

df

pd.to_datetime(df)

At least the year, the month and the day must be given here. One of the other three variables can be omitted at will. But month and day are not always numerical. The datetime library is ideal for these cases. Here are three examples of how differently formatted dates can be brought into a uniform format:

date_string1 = 'Wednesday, June 6, 2018'
date_string2 = '6/6/18'
date_string3 = '06-06-2018'


date_date1 = datetime.strptime(date_str1, '%A, %B %d, %Y')
date_date2 = datetime.strptime(date_str2, '%m/%d/%y')
date_date3 = datetime.strptime(date_str3, '%m-%d-%Y')


print(date_date1)
print(date_date2)
print(date_date3)

Here is a short list of the most common directives:

You can find the full list of directives “here”.

7 Conclusion

In this post we saw the different types of data and how to convert them to any other.