| 1 |
Then-elo is not available in the database as far as i can see, so it has to be recalculated or extracted from the replay files. The provided dataset does not contain any elo information, only lists of account ID's of winners and losers.
|
1 |
Then-elo is not available in the database as far as i can see, so it has to be recalculated or extracted from the replay files. The provided dataset does not contain any elo information, only lists of account ID's of winners and losers.
|
| 2 |
\n
|
2 |
\n
|
| 3 |
Extracting the information from replays is not currently possible because i can't find the replays from 2014. I think KR's data was obtained using the "Fixer" program, which basically recomputes elo from historic data.
|
3 |
Extracting the information from replays is not currently possible because i can't find the replays from 2014. I think KR's data was obtained using the "Fixer" program, which basically recomputes elo from historic data.
|
| 4 |
\n
|
4 |
\n
|
| 5 |
It is possible, though, to obtain elo [i]changes[/i] for each player in each battle.
|
5 |
It is possible, though, to obtain elo [i]changes[/i] for each player in each battle.
|
| 6 |
\n
|
6 |
\n
|
| 7 |
On
a
practica/technical
note,
you
can
use
a
local
database
to
store
intermediate
results,
so
it
won't
be
that
expensive.
|
7 |
On
a
practica/technical
note,
you
can
use
a
local
database
to
store
intermediate
results,
so
it
won't
be
that
expensive
to
keep
a
running
rating
for
everyone.
|
!predict is wrong! - New Prediction System for Teams