What Is API and How To Use Youtube API
Explaining the concept of API in plain English and how to scrape Youtube data using tuber package
Last
time I talked about strategies for a successful Youtuber. It was such
an exciting project for me. But It wasn’t an easy one because I had to
struggle with searching for how to get Youtube API key. There are lots
of resources already but most of them are for web developers, and
tutorials for R users are hardly seeable. After I managed to get the API
key spending a whole day, I came across with the idea why not I leave a
post for someone who would have the same problem with me.
Today
I’m going to talk about the concept of API and how to get API key. I
will also introduce tuber package for analyzing data in R. If you
haven’t used Google APIs and you are interested in doing a data science
project with them, this post will be a good start.
What is API?
If
you are unfamiliar with the word API, it could be hard to grasp the
meaning of it at first. API stands for Application Programming
Interface. Is this too ‘programmer’ language for you? That’s fine. I
will explain the concept step by step.
For your understanding, it would be better to start with two different interfaces. There are two kinds of interface, User Interface and Application Programming Interface.
To simply put, UI is the interface for common users while API is the
interface for programmers. You can see an example of UI below. It’s a
common web page we encounter every day. How do we use them? We can use
them simply by clicking the picture or words. This is the level for the
common people, the interface for users. How could it be possible? That’s
because the web developer hid all the codes required behind the button.
And that is the interface for programmers where the codes and commands
are living. In other words, API. You can think API as a set of codes,
protocols, and functions to work and interact with software
applications. With UI we work with simple buttons and tools while with
API we work with predefined functions.
Still
hard for you? Then let’s suppose you are sitting in a restaurant. You
want to have a nice beef steak and a glass of red wine for your dinner.
You sit on the table and call.. who? The waiters! Then you get the menu
and order what you want. After a few minutes later they will bring the
dish. This is what API does. You don’t need to roll up your sleeves and
do all the stuff like preparing the ingredients, cutting the vegetables,
and grilling the meat by yourself. Waiters will do these steps for you,
and all you have to do is just calling them and start your work from
that stand.
Likewise
when the developers are working with programming tools or when the data
scientists are wrangling with data, they don’t write every single code
every single time. They use pre-made functions which are the sets of
those basic codes operating under the hood. For instance, when you want
to join two data into one, you call the join function (a ‘join’ waiter)
and order it to work as you command (ordering the menu). Then it will
return the required results (your dish).
Why API?
API
now became a new kind of business models and strategies for companies
to equip in the big data era. Then what’s so special? What makes API so
hot these days? Let’s briefly talk about what we can do with it in a
business context.
APIs
can generate massive amounts of value both internally and eternally.
Managing and processing data is one of the crucial factors in business
management, and every company has built IT systems. As the size of the
data is increasing exponentially, however, there is a limitation to cope
with all the data through the traditional IT system. In this sense,
applying API can be a solution with better efficiency and security. It
can break down barriers between systems, which enables simplifying work
processes, inter-cooperation between organizations and higher protection
of data.
External
merits of API are even fancier. If a company open their API either
publicly or with extra fees, it can provide new services and acquire
potential customers to their side. Customers can experience a higher
level of services that weren’t available before. By offering API
services, third-party developers can build whole new kinds of products
that even the companies had never thought of. For example, Google Map,
the most popular API among developers, wasn’t expected to draw that much
effects at first. By Applying those data to real estate and various
other fields, developers brought much higher values and assets back to
Google.
Nowadays
the number of API is continuously increasing, and this trend will be
ongoing or even more. Now we can say that the implementation and
management of API of a company is one of the critical factors for its
competitive and strategic values.
So how to use API?
To
work with API, you need to get an authorized key first. It’s for
obtaining an authorized key to be connected with API. API Providers,
Youtube in this case, don’t merely provide their service without
control. To preserve this interface and manage the users, they offer a
unique access key to each user. With this, we can connect to the
interface of the application. Just like we connect to electricity by
plugging the power cord into an electric outlet, we plug our server into
the storage with this unique key. And from that point, we can use the
data and protocols of API.
So
now let’s start with how to get the authorized key. There is a very
nice tutorial for getting the key. If you’re a web developer, you can
follow this tutorial step by step.
If
you are to do some data analysis, you don’t need all these steps. You
follow the steps until number 9, and then choose ‘other’ instead of ‘web
application.’ There are two different types of clients on the developer
console, and the access methods are different for each case. Therefore
if you try to get the key by ‘web application’ and ask for Oauth from
your local computer, you could meet the error message like what I got.
So click ‘other’ option then you will get the client key and secret key. Copy and paste them on your local computer.
The tuber package for Youtube API
In R, there is tuber package
which is made only for Youtube data analysis. You can draw various data
including the number of videos of a particular channel, the number of
views, likes and comments on a video and so on. To call this data in R,
you need to install ‘tuber’ package first. Then you ask Google for the
authorization token.
install.package(tuber) library(tuber)
client_key = '' client_secret = '' yt_oauth(app_id = client_key, app_secret = client_secret)
When
you comment this code, a browser will pop up with some numbers allowing
you to authorize the application. If you check your local tool, R will
already be waiting for the numbers so copy and Paste the serial numbers.
Now you finally get to be connected with Youtube API.
I will also introduce some of the functions that could be useful for analyzing Youtube data from
tuber package.get_all_channel_video_stats: This is the function for getting statistics on all the videos in a channel. You need a channel_id to use this.get_stats: This gives you statistics of a video with video_id. The return values are the count of views, likes, dislikes, favorites, and comments.get_video_details: This provides more specific information as the publish date of a video. You can also get titles, descriptions, thumbnails, and categories.get_all_comments: It gives you all the comments for a video so has great usage for text mining in respect of particular topics or channelslist_channel_activities: This returns a list of channel activity.list_videos: It returns the most popular videos.
There are other useful API calls, so I recommend you to check them. You can browse them from here.
Resources
- An excellent video for what is API: https://www.youtube.com/watch?v=s7wmiS2mSXY&t=75s
- Harvard report on strategic values of API: https://hbr.org/2015/01/the-strategic-value-of-apis
- Want to study more about API? ProgrammableWeb is an ‘API’ university delivering almost everything about API: https://www.programmableweb.com/
- Don’t know what to do with Youtube data? Check out my previous work. You could get new inspirations from them!
There
are plenty of APIs already but, a small number of them open publicly.
As data will directly transfer to the competitiveness in today’s world, I
do understand companies’ intentions. I believe, however, they shouldn’t
neglect the potentials of sharing economy. By making their internal
data accessible, they will unleash the creativity of developers and data
scientists all over the world to devise whole new uses for the data.
Just like what we have done with open sources over the last years.
Thank
you for reading and hope you found this post helpful. If there is
something need to be corrected, please share your insight! If you’d like
to encourage an aspiring data scientist, please hit 👏 👏 👏! I’m
always open to hearing your thoughts so feel free to share or contact me
on LinkedIn. I’ll be back with another exciting story. Until then, happy machine learning.
https://towardsdatascience.com/what-is-api-and-how-to-use-youtube-api-65525744f520
Comments
Post a Comment