Data Analysis From Scratch With Python: Step By Step Guide

5. Overview & Objectives
Let’s set some expectations here so you know where you’re going. This is also to
introduce about the limitations of Python, data analysis, data science, and
machine learning (and also the key differences). Let’s start.
Data Analysis vs Data Science vs Machine Learning
Data Analysis and Data Science are almost the same because they share the
same goal, which is to derive insights from data and use it for better decision
making.
Often, data analysis is associated with using Microsoft Excel and other tools for
summarizing data and finding patterns. On the other hand, data science is often
associated with using programming to deal with massive data sets. In fact, data
science became popular as a result of the generation of gigabytes of data coming
from online sources and activities (search engines, social media).
Being a data scientist sounds way cooler than being a data analyst. Although the
job functions might be similar and overlapping, it all deals with discovering
patterns and generating insights from data. It’s also about asking intelligent
questions about the nature of the data (e.g. Are data points form organic clusters?
Is there really a connection between age and cancer?).
What about machine learning? Often, the terms data science and machine
learning are used interchangeably. That’s because the latter is about “learning
from data.” When applying machine learning algorithms, the computer detects
patterns and uses “what it learned” on new data.
For instance, we want to know if a person will pay his debts. Luckily we have a
sizable dataset about different people who either paid his debt or not. We also
have collected other data (creating customer profiles) such as age, income range,
location, and occupation. When we apply the appropriate machine learning
algorithm, the computer will learn from the data. We can then input new data
(new info from a new applicant) and what the computer learned will be applied
to that new data.
We might then create a simple program that immediately evaluates whether a
person will pay his debts or not based on his information (age, income range,
location, and occupation). This is an example of using data to predict someone’s

---

likely behavior.
Possibilities
Learning from data opens a lot of possibilities especially in predictions and
optimizations. This has become a reality thanks to availability of massive
datasets and superior computer processing power. We can now process data in
gigabytes within a day using computers or cloud capabilities.
Although data science and machine learning algorithms are still far from perfect,
these are already useful in many applications such as image recognition, product
recommendations, search engine rankings, and medical diagnosis. And to this
moment, scientists and engineers around the globe continue to improve the
accuracy and performance of their tools, models, and analysis.