Mamba Hiring: DIY statistics

Searching for a job

It happened here that I suddenly began to look for a new job. Interviews, tests, well, you yourself know everything. And one of the tests was the notorious Mamba Hiring, launched back in 2011.

Having passed the test and scored 217 points, I remembered a Facebook application in which I was tested from a year ago and where I had 242 points. Having looked and grieved at my numbers, I thought that again there was a rather average result. Oh? Is he such an average? We need to look at the statistics, maybe not everything is so bad? What is the “average” value in reality?

But there is no statistics anywhere, maybe I didn’t look well and ask to throw a link at me, but the maximum that I found is comparative statistics on your friends on Facebook and fragmentary outdated statistics from the links mentioned above.
')
But since the beginning of the test, a couple of years have passed and a lot of results have accumulated there. It’s a pity, of course, that there is no division by gender, age and all that, but is it interesting to see at least the big picture?

results

I am not a statistician and far enough from this topic, but I suspect that the minimum score is still not ten parrots. Ten parrots are those who just clicked at random all the way forward, but still ...

Minimum result: 10
Maximum result: 497
Medium Result: 166

Number of tested at the time of this writing (May): 83591
Number tested at the time of publication of the article: 86354

Results graph (clickable, leads to Google Drive with an interactive option):
Chart option from eyeless_watcher :

What about the process?

Everything is just like a nut. When we finish the test on the site corp.mamba.ru/test , we are given a link to a graphical widget of the form "http://www.corp.mamba.ru/test/widget.phtml?id= 9000 " It is interesting to us with its ID value . If you change it, you can see another, alien result.

The algorithm was born instantly.
First run: let's sculpt the loop to iterate through ID'shnikov, download the picture, put in the daddy.
2nd run: we go through the pictures, cut them to the area with the number, put them into the black box of text recognition in the person of TesseractOCR , and dump the resulting value into the database.

The code of scripts on the githaba - just please without holivars, the whole thing is written in 5-10 minutes and at one time.

In conclusion, I would like to ask Mambu to post the results from testing through the Facebook application, I don’t think that having the data on the social account of the tester can be very difficult to compare the specific result with a specific age and sex :)

Yes, by the way, who needs the source data, then they are a .sql dump. I can even ~ 83,500 (~ 700 mb) of pictures to post if someone needs it.