Statistical analysis of patch 12.3

The dataser used to this analysis is in (kaggle) [https://www.kaggle.com/vivovinco/league-of-legends-champion-stats] and the regions of each champion was manually inserted. This analysis was made in R

Statistics

The first step is to realize the statisticals tests to know with what kind of distribution we are dealing. Our variable of interest is the win rate, which will lead us in the comparations between the others variables. To dummies: The statistical analysis has to respect some rules. To know what rules will guide us we’ll check the normality of the data.

Normality Tests: Shapiro-Wilk and Kolmogorov-Smirnov

## 
##  Shapiro-Wilk normality test
## 
## data:  lol12_3$win
## W = 0.99481, p-value = 0.5952
## 
##  One-sample Kolmogorov-Smirnov test
## 
## data:  lol12_3$win
## D = 0.040485, p-value = 0.8149
## alternative hypothesis: two-sided

Both showed us that win_rate is normal (p-valor>0.05). So we are going to apply ANOVA and Tukey to check is Region and Tier has something to do with the Win Rate distribution.

##                 Df Sum Sq Mean Sq F value Pr(>F)  
## lol12_3$region  13   67.8   5.218     1.8 0.0442 *
## Residuals      224  649.5   2.900                 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##               Df Sum Sq Mean Sq F value   Pr(>F)    
## lol12_3$tier   5  162.0   32.41   13.54 1.35e-11 ***
## Residuals    232  555.3    2.39                     
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Both p < 0.05, in this case showing that selecting a champion by it’s Tier and/or it’s Region could raise your winning probability in patch 12.3 We could see that Tier was more important than Region. Now let’s see what Tier and Region the champion must belong so you can use it as a filter when drafting.

Everything that the pointed line(in vertical) touches means that they are statistical equals. The most centralized the horizontal line ir in the pointed one, more the compared Tiers are similar. We can see that God Tier and A are not very similar, while B is more alike God than A itself. That’s curious, let’s keep exploring

With a correlational matrix we’ll observe how the others quantitatives metrics, How they related between themselves.

The understading is simple, the more the correlation gets far from 0(to +1 or -1), more intense are these correlations.

Ploting

Now we enter in a section that is more visual and less tecnical. Next wi have a bunch of graphics that will ilustrate how was the patch, by the opticas of geography and Tier

Something interesting happend here. We saw that Tier A was not very alike God, S or B (that are very similar), but now we learned why: A had a better performance than they.In excepetion of it’s minimun quantile, all other was higher than all of the other’s ones.

Well, now let’s see which regions are the bests, and what are the worse. But a mere whim, let’s be a little bit more minuncious. The next graphic has the pick rate implicit. It is represented by the circle’s size, the higher the pick rate, bigger the circle.

Above we saw how was the performance of the champions separated by Region, Below we’ll see a very similar graphic, but separeted by Tier.

Approaching the end of this analysis, the Regions’ performance in Summoner’s Rift.

With all these informations in hands, you can take more decisions more rationally to improve your performance. Remember: I just gave you the informations, the choice maker is YOU!