Data Analysis: How to Access HotS Data

General Discussion
1

Hi,
I’d like to find out if the loss of the early teamfight increases the likelihood of losing the whole match, i. e. I want to address the issue of people saying “gg” when they lose the first fight.

So, I know that these data can be accessed somehow (as HotsLogs can do it). I’m interested in

  • no. of friendly deaths in the first minute
  • no. of enemy deaths in the first minute
  • outcome of the match

Can someone point me to the right direction?

I’m looking to perform a simple logistic regression and will share the results afterwards.

Thanks.

2

That’s a lot of work buddy. You gotta parse chat logs for the “gg” then figure out if its a real “GG?” or a sarcastic “gg” LOL. Then! figure out if the fights as you said… early vs late. Then!!! try to figure out the advantages the winning team had (early game strength vs late game heroes). Then see if there is a ZAG player who just lanes all game and ignores everyone else. Then! parse out all the games where someone was AFK for the first 2 or 3 minutes of the game. Might wanna exclude any “hi guise I’m drunk!” … cause thats a factor too. Like it never ends. No control for the data.

Like good luck! LOL

P. S. Look up HOTS APIs… in the regular battlenet forum.

1 Like
3

Seeking the association between losing the early fight (and the volume of the loss measured by the K/D) and the outcome of the match. I’m not answering the question “Does saying gg after a lost fight associate with a higher chance of losing” but “Does losing the early fight associate with a higher chance of losing”. I.e. I’m trying to find out if the first loss predicts the outcome

4

Check out the "API’ right at the bottom of this very webpage.
P. S. Can’t say if it will give you exactly what you need but I know its a way to access data especially for the other games. As an alternative you might want to try and figure out how to use data collected by well known HOTS websites… like hotslogs (currently run by Zam network).

5

Thanks! I don’t know how to work that though. I’m a statistician not a programmer lol

6

Yea sorry its not totally straightforward. Maybe reach out to hotslogs. The APIs are mostly for making apps and websites using the data Blizzard provides.

7

I would appreciate if this thread reached someone who would just email me the data in a .csv format or something

But I’m too optimistic, I think :slight_smile:
In the meantime, I’m at least collecting the data from my own games

1 Like
8

You could probably shift through hotslogs for game length data but losing a teamfight early affects the mentality of a team more than anything. It provides about as much experience as a minion wave so it doesn’t affect the game that much. Its just because so idiot thinks a game is over is the only way it really affects a team.

9

And I’m going to analyse that precisely.

In my experience, losing the first fight means nothing.

10

I would assume that the better team will usually have more early game kills and is more likely to win the match in general. You might be able to look at pro tournaments/games if nothing else.

11

it sounds logical, but I don’t want to trust the intuition, I want to go by the data – the early win might just be
a) a fluke
b) an effect of your early game comp

EDIT: And good idea with the pro data! However, it is a question if those game represent our reality.

12

In all honesty you should really learn how to, at the very least, pull API. Scripting and programming are very different things.

13

Is there a guide you would suggest?

14

Generically speaking, codeacademy is pretty good in terms of easy and free access.

15

Ok, thanks for the suggestion

:slight_smile:

16

No worries :slight_smile:

Usually, my favorite folks in a company are the data people. Not because they house the secrets, but rather because they tend to have a pretty broad range of personal skills that they use relative to data science. Scripting, html to whatever degree, various programming languages, etc. And they get paid pretty darn well.

(Point being is the future is bright for you once you have begun your own personal expansion of skills that run parallel to data.)

1 Like