It's tough. Here's where I'm at personally with League data:
A lot of things are instantly useless. So for example, the winrate of X champion at Y tournament is pretty much always bad as an analysis tool. Not only are there usually not enough data points to do anything more than, "yeah this champion is probably strong on average," you also have biased data by, say, EDG playing more Lucian+Nami than any other team. Now you're just measuring EDG's win rate, not Lucian's.
This trend continues with things like, "Show my Viper's CSD." Well, we already established that EDG plays a lot of Lucian Nami, so he probably has inflated CSD from lane matchups, not necessarily because of player skill.
You can build upon that and construct a model for, "OK, what is the average Lucian vs. Ashe CSD and how does that compare here?" You can maybe construct Lucian vs. Ashe across all pro play for an entire year but more likely you're looking at Master+ solo queue instead. Keep in...
Read more