While many public datasets (on Kaggle and the like) provide Apple App Store data, I was surprised to find that there are not many counterpart datasets available for Google Play Store apps anywhere on the web. On digging deeper, I found out that iTunes App Store webpage deploys a nicely indexed appendix-like structure to allow for simple and easy web scraping. On the other hand, Google Play Store uses sophisticated modern-day techniques (like dynamic page load) using JQuery making scraping more challenging.
Following this, I published my dataset on Kaggle (you can find it here) to drive app-making businesses to success. Actionable insights can be drawn for developers to work on and capture the Android market!
Methodology:
For my web-scraping project, I used Selenium to scrape all the apps on Play Store. The apps on Play Store are divided into about 33 categories. To get apps in all these categories, I wrote a Python script which iteratively loops through all categories, clicks on one category at a time, scrapes all the apps, returns to the main page (dropdown as shown below) and then proceeds to the next category.
Category selection option on Google Play Store
However, this approach was not sufficient since Google uses its own intelligent recommendation algorithms and lists only a few selected apps for each category, based on your user history till now. Due to this reason I was able to scrape only about ~3.3k apps, which would definitely underrepresent the Android market.
To overcome this, I adopted a workaround in which I used the 'search bar' on the top of the page. I made 26Cr where r ∈ [1,5] permutations of 26 alphabetsto get a list of all possible substrings (of length 1 to r). I triggered a new search using each of these substrings to scrape new apps. Through this approach, I was able to get another 10.5k apps!
Dataset Structure:
From each individual app page, I scraped the following details:
App name
Category
Rating (on a scale of 5)
Number of Reviews
Size of the app
Number of downloads
App type - Free or Paid
Price
Content Rating
Genres
Last Updated On
Current Version
Required Android Version
Under the 'Reviews' tab for each app, I scraped the following details:
Reviewer name
Review text
Review date
Review rating
Number of likes (on the review)
I stored these two datasets in separate .csv files. 'App name' was used as the join column for review analysis on each app.
Data Analysis:
A quick look at the Android market distribution
Family and Game apps have the highest market prevalance.
Interestingly, Tools, Business and Medical apps are also catching up.
Average rating of apps
Do any apps perform really good or really bad?
Average app rating = 4.173243045387998
Best Performing Categories
Almost all app categories perform decently. Health and Fitness and Books and Reference produce the highest quality apps with 50% apps having a rating greater than 4.5. This is extremely high!
On the contrary, 50% of apps in the Dating category have a rating lesser than the average rating
A few junk apps also exist in the Lifestyle, Family and Finance category.
Sizing Strategy - Light Vs Bulky?
How do app sizes impact the app rating?
Most top rated apps are optimally sized between ~2MB to ~40MB - neither too light nor too heavy.
Most bulky apps ( >50MB) belong to the Game and Family category. Despite this, these bulky apps are fairly highly rated indicating that they are bulky for a purpose.
Pricing Strategy - Free Vs Paid?
How do app prices impact app rating?
Most top rated apps are optimally priced between ~1$ to ~30$. There are only a very few apps priced above 20$.
Clearly, Medical and Family apps are the most expensive. Some medical apps extend even upto 80$
All other apps are priced under 30$
Surprisingly, all game apps are reasonably priced below 20$.
Distribution of free and paid apps across categories
Are paid apps downloaded as much as free apps?
Paid apps have a relatively lower number of downloads than free apps. However, it is not too bad.
How do the sizes of paid apps vary?
Majority of the paid apps that are highly rated have small sizes. This means that most paid apps are designed and developed to cater to specific functionalities and hence are not bulky.
Users prefer to pay for apps that are light-weighted. A paid app that is bulky may not perform well in the market.
Basic sentiment analysis - User reviews
Health and Fitness apps perform the best, having more than 85% positive reviews.
On the contrary, many Game and Social apps perform bad leading to 50% positive and 50% negative.
Free apps receive a lot of harsh comments which are indicated as outliers on the negative Y-axis.
Users are more lenient and tolerant while reviewing paid apps - moderate choice of words. They are never extremely negative while reviewing a paid app.
WORDCLOUD - A quick look on reviews
Paid apps
Free apps
Key Insights:
Average rating of (active) apps on Google Play Store is 4.17.
Users prefer to pay for apps that are light-weighted. Thus, a paid app that is bulky may not perform well in the market.
Most of the top rated apps are optimally sized between ~2MB to ~40MB - neither too light nor too heavy.
Most of the top rated apps are optimally priced between ~1$ to ~30$ - neither too cheap nor too expensive.
Medical and Family apps are the most expensive and even extend upto 80$.
Users tend to download a given app more if it has been reviewed by a large number of people.
Health and Fitness apps receive more than 85% positive reviews. Game and Social apps receive mixed feedback - 50% positive and 50% negative.
Users are more critical and harsh while reviewing free apps. They are never extremely negative while reviewing a paid app.
Google
Every once inside a when we select blogs that we read. Listed below would be the most up-to-date websites that we pick.
Google January 8, 2021
Google
The info mentioned inside the report are a few of the most effective offered.
CBD For Dogs December 18, 2020
CBD For Dogs
[...]usually posts some incredibly interesting stuff like this. If you are new to this site[...]
Google December 3, 2020
Google
Here are some links to web pages that we link to because we feel they're worth visiting.
RDP August 28, 2020
RDP
[...]we came across a cool web site that you might enjoy. Take a search when you want[...]
MKsOrb August 26, 2020
MKsOrb
[...]Wonderful story, reckoned we could combine several unrelated information, nevertheless actually worth taking a search, whoa did one discover about Mid East has got much more problerms as well [...]
onhaxx.me August 19, 2020
onhaxx.me
[...]Every the moment in a even though we opt for blogs that we study. Listed below are the latest web sites that we select [...]
mksorb.com August 5, 2020
mksorb.com
[...]that would be the end of this article. Here youll locate some sites that we feel youll enjoy, just click the hyperlinks over[...]
mksorb.com July 30, 2020
mksorb.com
[...]just beneath, are a lot of absolutely not related web sites to ours, having said that, they may be certainly really worth going over[...]
cbd cats July 9, 2020
cbd cats
[...]the time to study or stop by the content material or web sites we have linked to beneath the[...]
It will be a great delight to stay in touch with you. [email protected]
Vijay December 21, 2018
Hello Lavanya, it is indeed a great work and research carried on by you for scraping and analyzing Google Play sore Apps. Will you be comfortable shedding a bit more light on the logic of the code you worked on by sharing it with me? It would be a great help for me, as I am still learning and want to pursue this path. Thank you so much again, for sharing your valuable research with the world.
Milan November 28, 2018
Hello Lavanya,
Really great work at scraping and analyzing Play Store Apps. Very impressive stuff.
Would it be possible to share the web-scraper used to get data from play store? It would be really helpful to me for learning purposes.
Thank you so much.
Milan
[email protected]