Analysis of programming languages tags used on Steem

The purpose of the following analysis is to investigate the popularity of programming tags and their development over time.

Outline

• Most popular programming tags
• Development over time
• Average post payout

Scope of Analysis

The data was collected from programming posts from the very beginning of Steem blockchain up to the present.

Tools

• SteemSQL
• Python 3.6 (with matplotlib library)

Results

Most popular programming tags

The first step was to delete all posts that are not related to programming and leaving only those columns that will be useful in further analysis:

• creation date
• url
• payout
• tags

2017-06-30  /programming/@profitgenerator/learn-basic-python-programming-ep-4-let-s-build-a-calculator  5.9480  programming,howto,education,python,tutorial
2017-07-01  /programming/@qed/installing-haskell-idris-and-atom-idris-2 0.0560  programming,technology,math,video,steemit
2017-06-25  /code/@cosmobug/what-can-you-do-with-code   0.0000  code,programming,python,csharp,visualstudio
...

Using a simple script I checked the occurence of tags related to programming.

counter = Counter()
with open('programming.tsv', 'r', encoding='utf8') as f:
    for line in f:
        columns = line.strip().split('\t')
        tags = columns[3].split(',')
        for tag in tags:
            counter[tag] += 1
for tag, c in sorted(counter.items(), key=itemgetter(1), reverse=True):
    print(tag, c)

Some tags such as python, rust, c, r do not necessarily concern programming, so I started by finding general programming tags so that I could use them for filtering.

Tag	Occurrence
programming	10200
technology	2503
coding	1774
utopian-io	1217
tutorial	913
steemdev	544
development	385
computer	353
code	280
opensource	260
software	256
tech	241
learning	225
tutorials	217
html	193
linux	188
ai	159
web	158
design	147
dev	143
security	139
android	135
developer	130
steemiteducation	125
learn	114
computers	110
hacking	95
css	91
machine-learning	91
deep-learning	85

The next step was to count programming languages tags. Some tags refer to the same language (rust and rust-lang, go and golang, cpp and cplusplus and c-plusplus), so you need to take this into account.

counter = Counter()
with open('programming.tsv', 'r', encoding='utf8') as f:
    for line in f:
        columns = line.strip().split('\t')
        tags = (columns[3]
            .replace('rust-lang', 'rust')
            .replace('golang', 'go')
            .replace('cplusplus', 'cpp')
            .replace('c-plusplus', 'cpp')
            .split(','))
        tags = set(tags)
        if tags & set(common_tags):
            for tag in tags:
                counter[tag] += 1
for tag, c in sorted(counter.items(), key=itemgetter(1), reverse=True):
    if tag not in common_tags:
        print(tag, c)

The result looks as follows.

.	Tag	Occurrence
1	python	1013
2	java	595
3	javascript	552
4	php	193
5	c	114
6	cpp	97
7	csharp	52
8	go	50
9	solidity	49
10	ruby	41
11	rust	40
12	kotlin	37
13	r	28
14	scratch	19
15	mysql	18
16	assembly	17
17	elixir	17
18	swift	15
19	lua	11
20	bash	10

I also checked the 20 most popular tags from https://stackoverflow.com/tags to be able to compare the results.

.	Tag	Occurence
1	javascript	1553961
2	java	1370330
3	c#	1177662
4	php	1165762
5	python	891863
6	c++	553889
7	mysql	504295
8	objective-c	282490
9	c	270118
10	r	221396
11	ruby	191670
12	swift	180026
13	vb.net	115839
14	bash	95125
15	vba	93824
16	postgresql	80043
17	matlab	76946
18	scala	75314
19	perl	58632
20	delphi	40852

Languages ​​such as python,java, javascript, php occupy top positions in both tables. Relatively low in the Steem table is csharp.

Below is the list of languages only included in the Steem ranking:

• go
• solidity
• rust
• kotlin
• scratch
• assembly
• elixir
• lua

And the languages only included in the stackoverflow ranking:

• objective-c
• vb.net
• vba
• postgresql
• matlab
• scala
• perl
• delphi

The first factor that can affect these differences is the fact that the Steem blockchain works much shorter and some of the languages were popular some time ago: objective-c, which is replaced by swift; perl, which is replaced by, for example, python and delphi, which is now very rarely used. The Steem ranking also did not include both languages ​​from the VisualBasic family: VisualBasic.Net (vb.net) and VisualBasic for Application (vba). The low popularity of the postgresql tag is probably due to the fact that the most popular Steem blockchain databases use other solutions: SteemSQL - mssql, sbds - mysql, SteemData - mongodb. Personally, I am surprised by the low popularity of scala tag in Steem.

The reason why the languages ​​rust,kotlin, elixir are quite high in the Steem ranking compared to stackoverflow is the fact that they are relatively new technologies. Scratch is a visual programming language, especially for children and youth, so it's not surprising that it's not popular at Stackoverflow, which is designed for professional programmers. Solidity is a programming language for the Ethereum platform, hence its high position in the Steem ranking.

Other languages ​​used by platforms similar to Ethereum:

• Lisk: javascript
• Cardano: haskell
• NEO: csharp, vb.net, fsharp, java, kotlin, python
• EOS: wren
• Stratis: csharp

Let's also look at the popularity of tags on the pie chart.

Development over time

The following chart shows the number of posts related to a top 10 most popular programming languages in given months.
Popularity of which tags will grow fastest? I think that it will be the top 4: python, javascript, java, php because they are they generally very popular, and used by steem related libraries:

• steem-python
• steem-js
• steemJ
• steem-php

I think that the solidity can also count on growth due to the development of the Ethereum platform.

Below is the chart for the remaining tags. The division into two charts was intended to increase readability. Among the languages ​​of the second tenth, the greatest potential for growth is for the relatively new ones, ie rust, kotlin, elixir. And probably r because of machine learning popularity.

Average post payout

The average payout depending on the tag is quite diverse. More popular tags generate higher payouts, probably because of reaching a wider audience. It should also be taken into account that the result for less common tags may be somewhat biased due to the too small sample from which the result was determined.

---

Posted on Utopian.io - Rewarding Open Source Contributors

---