Beginner’s guide: My start in Data Science

A beginner’s guide for beginners

This article was originally posted here

Before The Outset

I’ve never been good enough to make complex manipulations on data but writing algorithms and building web applications is a big Yes.

Notes:

→ This is not a guide to becoming a full Data Science Engineer, I’m just sharing what I started with in this field.

→ This is not a unique path either.

Instead, you can consider it as a start study plan for those coming from Software Development and want to be Junior Data Science engineer.

The voice of a beginner for others beginners. :)

What is Data Science?

Data science is a multidisciplinary blend of data inference, algorithm development, and technology in order to solve analytically complex problems.

What is Data Science?

For me, it is all the things we are doing with data that can solve some problems and come out with business value or growth.

Kind of Data

As we said up there it is about manipulating data the whole time, but what kind of data can we manipulate?

On the internet or in large enterprise applications there is a lot of data coming from different sources such as social media, calls to action, simple forms, log data, transactions, emails …
All things we are doing online or in others place require the most times that we input data, these data can be in different types:

• Text
• Photo
• Audio
• Video
• …

Also in different formats:

• Structured data: those with a certain degree of organization for further querying and/or analysis. As the one stored in Relational Database Management System or in Json, Xml, Xls files.
• Semi-structured and Unstructured data: easy to understand (not formatted), the opposite of the first.

Programming Languages for Data Science

There are a lot of languages for Data Sciencing (hahaha), and some of them are very popular and more used than others.
There are a few: R, Python, Java …

Job Posting by Indeed

Job Seekers by Indeed

Google Trend

Percentage of search interest in R and Python for Data Science: R in Red and Python in sky Blue

Why I choose to start Data Science with Python

I have not just chosen to learn Data Science because it is also better paid in the tech industry. I’m firstly a passionate developer, OK? Then love to implement and discover a lot and I will surely use it in a venture we launch with some partners.

Python because it is also used a lot in this field and is one of the first programming languages I started with and found easy.

Intro to Python for Data Science

I started Data Science with a free and very educative certification available on www.datacamp.com, it helped me to introduce myself to this field to be able to start in a new way in my career.

The course

Learn Python for Data Science - Online Course

What I learned there

• Manipulating Python list in deep
• Manipulating Numpy array
• Subsetting Numpy array
• Subsetting 2D Numpy array
• Simple exploration of data
• Basic statistic

Practice

There are a lot of exercises and XP I earned there on this free course, I will show you some examples of things I learned there, but impossible to put all here, it is not the intent of this post and it is too much:

• 4700 XP Earned
• 1 Courses Completed
• 57 Exercises Aced

1. Python List

• Create a list:

mySimpleList = [12, 43, 54, 34, 90] #Simple list with same type
myWeirdList  = ['a', 43, 54, 'c', 90]   #Different types of items

• Print a list

print(myList) # Knowing that the list is already created

• List of list

countries = [["Cameroon", "CM"],
         ["Nigeria", "NG"],
         ["France", "FR"],
         ["Gabon", "GA"]]

• Type of a variable

To print the type of variable, just hit this:

print(type(myVariable))

• Index

list[4] = list[-2] = 5

• Subsetting list

A subsetting always returns a list. Here the first index is included in the result and the last is not.

list[1:4] = [2, 3, 4] # From index 1 to index 3 included
list[:4] = [1, 2, 3, 4] # From the start to index 3 included
list[1:] = [2, 3, 4, 5, 6] # From index 1 to the end

2. Numpy Array

• Install Numpy: pip3 install numpy
• Import Numpy

There is some manner to import python package/function, let’s focus on these two for this post:

import numpy # Here we will address numpy array with numpy.array
import numpy as np # Address numpy array with np.array

• From list to numpy Array

# Using countries list declared up there
countries_np_array = np.array(countries)

• subsetting np array

countries_np_array[:, 1] = array([‘CM’, ‘NG’, ‘FR’, ‘GA’])
# Return all country code, all rows and the second column

age_array = np.array([2, 4, 6, 8])
age_selector = age_array >= 4 
# result array([False,  True,  True,  True], dtype=bool)
# Now use this selector to index the new array
print(age_array[age_selector])
# result array([4, 6, 8])

Note: Numpy does not allow multiple types on an array and will force all types to be the same.

age_array = np.array([True, 4, False, 8])
print(age_array)
#result array([1, 4, 0, 8])

• Operation over collections

age = [2, 4, 6, 8]
div = [2, 2, 2, 4]
age_array = np.array(age) # Numpy array of ages
div_array = np.array(div) # Numpy array of divs
print(age/div)  # divide python list
# Traceback (most recent call last):
# File “<stdin>”, line 1, in <module>
#TypeError: unsupported operand type(s) for /: ‘list’ and ‘list’
print(age_array/div_array) # Will compute without issue on each item
array([ 1.,  2.,  3.,  2.]) # result

Data Science deals with a lot of information to analyze, sort, and do other things on, then it needs to do mathematical operations over collections quickly.

3. Little stats with Numpy

• Average: np.mean(your_numpy_array_or_axis)
• Median: np.median(your_numpy_array_or_axis)
• Standard Deviation: np.std(your_numpy_array_or_axis)

Supposed we have an array representing the grade of 3 students of a class in two courses(French and English):

import numpy as np
student_grades = np.array([[12, 16], [15.5, 9], [5, 16]])
# Average of student's grade in French
# Here we select all the rows and the french axis(the first column)
french_average = np.average(student_grades[:, 0])
print(french_average)
# result: 10.833333333333334
# Standart deviation of student's grade in English
# We select all the rows and the english axis(the second column)
english_std = np.std(student_grades[:, 1])
print(english_std)
# result: 3.2998316455372216

Got my first certification in this field :)

There are also a lot of interesting community courses available but non-certifying:

Free Data Science and Analysis Training Courses | DataCamp

Let’s Work together

Ping me if you are looking, or if you would like to have a partner to study and master Data Science with, I’m available for peer learning.

Conclusion

I think that starting by practicing using a simple and detailed course with a good scope is worth it for learning new things.

Thrilled to have it and excited to learn more using online resources and other posts about Data Science. I would also like to know how you started, or which advice can you give to a beginner like me.

Thanks for reading this post, recommend and share if you enjoyed it.

Follow me on Twitter and LinkedIn, and visit my blog.

Cheers!