Getting Started¶
pyETT: Python library for Eleven VR Table Tennis data.
[1]:
from pyETT import ett
import pandas as pd
Player and user search¶
In order to get player’s information we first need to uniquely identify a player. We can use the user_search() method of an object of the class ETT to search players by username, as follows:
[2]:
eleven = ett.ETT()
eleven.user_search_dataframe('Highlander').head()
[2]:
| id | name | elo | rank | wins | losses | last_online | |
|---|---|---|---|---|---|---|---|
| 0 | 248681 | Highlandersge | 1488.0 | 361225 | 0 | 1 | 2021-03-30T17:29:28.963Z |
| 1 | 246892 | mwwhighlander | 1500.0 | 222424 | 0 | 0 | 2021-01-02T23:00:36.762Z |
| 2 | 218318 | .::Highlander::. | 1500.0 | 179327 | 0 | 0 | 2021-01-07T00:52:31.209Z |
| 3 | 166876 | The Highlander | 1530.0 | 16962 | 4 | 1 | 2020-12-01T20:22:39.752Z |
| 4 | 384291 | highlander10 | 1500.0 | 71453 | 0 | 0 | 2021-04-16T22:18:26.874Z |
Here, the function returns all players that contain the input username string ‘Highlander’. To search for a perfect match, we can pass the argument perfect_match=True:
[3]:
eleven.user_search_dataframe('highlanderNJ', perfect_match=True)
[3]:
| id | name | elo | rank | wins | losses | last_online | |
|---|---|---|---|---|---|---|---|
| 0 | 348353 | highlanderNJ | 2013.5 | 2042 | 475 | 372 | 2021-06-30T01:53:32.742Z |
In this search, we were able to find the player with username ‘highlanderNJ’ which turns out to be me. We can now instantiate a Player object to obtain his information:
[4]:
print(p:=ett.Player(348353))
highlanderNJ
[5]:
print(f"username:{p.name}\n elo:{p.elo}\n rank:{p.rank}\n wins:{p.wins}\n losses:{p.losses}\n")
username:highlanderNJ
elo:2013.5
rank:2042
wins:475
losses:372
We can see I have an Elo number of 2013.5, which puts me ranked in number 2042 in the world with 475 ranked wins and 372 losses. That’s not entirely bad as the total number of users around 500,000 players!
Matches and rounds¶
Now that we have instantiated a Player object, we can obtain his matches using the method get_matches() or get_matches_dataframe() to obtain the data in a nice pandas dataframe. These methods return ranked matches by default but you can pass the parameters unranked=True to obtain unranked matches.
[6]:
m = p.get_matches_dataframe()
m.head()
[6]:
| created_at | id | ranked | number_of_rounds | state | winning_team | losing_team | home_score | away_score | home_player | away_player | rounds | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 2021-06-30T01:34:17.740Z | 9837423 | True | 3 | 1 | 0 | 1 | 2 | 1 | highlanderNJ | RandomPerson0504 | [<pyETT.ett.Match.Round object at 0x7fcc741398... |
| 1 | 2021-06-24T15:19:50.324Z | 9652764 | True | 3 | 1 | 0 | 1 | 2 | 0 | highlanderNJ | ZombieEngineerz8 | [<pyETT.ett.Match.Round object at 0x7fcc3cdcfe... |
| 2 | 2021-06-24T15:09:48.858Z | 9652464 | True | 3 | 1 | 0 | 1 | 2 | 0 | highlanderNJ | mcstives | [<pyETT.ett.Match.Round object at 0x7fcc3cdcf7... |
| 3 | 2021-06-23T22:33:16.914Z | 9631520 | True | 3 | 1 | 0 | 1 | 2 | 1 | ilyap | highlanderNJ | [<pyETT.ett.Match.Round object at 0x7fcc3ce160... |
| 4 | 2021-06-23T22:26:44.411Z | 9631342 | True | 3 | 1 | 0 | 1 | 2 | 0 | LearnerForever | highlanderNJ | [<pyETT.ett.Match.Round object at 0x7fcc3ce164... |
We can see that my latest ranked match was against the player with username ‘RandomPerson0504’ and I won the match 2 x 1. In order to obtain the rounds information, we can use the method get_rounds_dataframe from the class Match, as follows:
[7]:
ett.Match.get_rounds_dataframe(m.rounds[0])
[7]:
| id | round_number | state | away_score | home_score | winner | created_at | |
|---|---|---|---|---|---|---|---|
| 0 | 22061114 | 0 | 0 | 8 | 11 | 0 | 2021-06-30T01:37:16.546Z |
| 1 | 22061034 | 0 | 0 | 11 | 7 | 1 | 2021-06-30T01:35:59.004Z |
| 2 | 22060923 | 0 | 0 | 9 | 11 | 0 | 2021-06-30T01:34:17.740Z |
Friends list, leaderboard and elo history¶
From the library, we can also obtain a player’s friends list with the method get_friends() from an object of the Class Player, which also has a corresponding method that returns a list of friends in a pandas dataframe:
[8]:
p.get_friends_dataframe().head()
[8]:
| id | name | elo | rank | wins | losses | last_online | |
|---|---|---|---|---|---|---|---|
| 0 | 35497 | perseph | 1626.0 | 13492 | 1208 | 1883 | 2021-06-20T22:50:59.471Z |
| 1 | 431729 | JoeSixPaK | 1804.5 | 5542 | 389 | 654 | 2021-06-30T02:15:30.270Z |
| 2 | 495551 | 77696c6c | 1500.0 | 165378 | 0 | 0 | 2021-06-14T16:21:31.027Z |
| 3 | 456082 | pete_ed5 | 1796.0 | 5807 | 123 | 128 | 2021-06-29T22:45:05.229Z |
| 4 | 249045 | Buttersmier | 1897.0 | 3601 | 80 | 74 | 2021-06-28T21:37:01.090Z |
I’m interested in comparing my Elo score with my friends’. For that purpose, we can get a player’s elo history with the method get_elo_history():
[9]:
p_elo = p.get_elo_history()
p_elo.head()
[9]:
| elo | |
|---|---|
| 2021-02-20T21:13:12.497Z | 1500.0 |
| 2021-02-24T03:06:30.809Z | 1513.0 |
| 2021-02-24T03:13:06.512Z | 1502.0 |
| 2021-02-24T03:16:20.774Z | 1489.0 |
| 2021-02-24T03:19:46.168Z | 1482.0 |
Let’s now pull the Elo history from 5 of my friends and put it all together in a dataframe including my own elo history and then plot them all:
[10]:
from functools import reduce
n_friends = 5
friends_elo = [f.get_elo_history().rename(columns={"elo":f.name}) for f in
p.get_friends()[-n_friends:] if f.get_elo_history() is not None]
friends_elo.append(p.get_elo_history().rename(columns={"elo":p.name}))
group_elo_df = reduce(lambda df1,df2: pd.merge(df1,df2,how='outer', left_index=True, right_index=True),
friends_elo).ffill(axis = 0)
[11]:
group_elo_df
[11]:
| Konowl77 | belmont | Plipplopplip | olofaa | Kennatius | highlanderNJ | |
|---|---|---|---|---|---|---|
| 2020-12-31T01:14:04.984Z | 1500.0 | NaN | NaN | NaN | NaN | NaN |
| 2020-12-31T18:45:27.553Z | 1493.0 | NaN | NaN | NaN | NaN | NaN |
| 2020-12-31T18:50:46.088Z | 1476.0 | NaN | NaN | NaN | NaN | NaN |
| 2021-01-01T00:34:21.158Z | 1468.0 | NaN | NaN | NaN | NaN | NaN |
| 2021-01-01T02:38:08.062Z | 1453.0 | NaN | NaN | NaN | NaN | NaN |
| ... | ... | ... | ... | ... | ... | ... |
| 2021-06-30T01:34:17.740Z | 1763.8 | 1907.2 | 1728.0 | 1506.0 | 1652.0 | 2005.5 |
| 2021-06-30T01:37:59.877Z | 1763.8 | 1907.2 | 1733.0 | 1506.0 | 1652.0 | 2005.5 |
| 2021-06-30T01:44:09.865Z | 1763.8 | 1907.2 | 1733.0 | 1506.0 | 1636.0 | 2005.5 |
| 2021-06-30T01:44:57.569Z | 1763.8 | 1907.2 | 1736.0 | 1506.0 | 1636.0 | 2005.5 |
| 2021-06-30T01:49:24.921Z | 1763.8 | 1907.2 | 1749.0 | 1506.0 | 1636.0 | 2005.5 |
4852 rows × 6 columns
[12]:
group_elo_df.plot(figsize=(10,8))
[12]:
<matplotlib.axes._subplots.AxesSubplot at 0x7fcc36dcd850>
Overall, my friends are making some good progress. Everyone seems to be improving over time.
Now let’s take a look at the leaderboard by using the method get_leaderboard_dataframe() from our eleven object. By default, this method returns the top 10 players. You can pas
[13]:
eleven.get_leaderboard_dataframe()
https://www.elevenvr.club/leaderboards/?start=0
https://www.elevenvr.club/leaderboards/?start=10
[13]:
| id | name | elo | rank | wins | losses | last_online | |
|---|---|---|---|---|---|---|---|
| 0 | 385684 | yedat | 3258.2 | 0 | 967 | 258 | 2021-06-28T19:21:18.464Z |
| 1 | 347650 | PROTT_VR | 3095.0 | 1 | 662 | 119 | 2021-06-22T19:37:25.710Z |
| 2 | 234850 | lapoctopus90 | 3087.1 | 2 | 712 | 246 | 2021-06-28T17:55:13.346Z |
| 3 | 289382 | gselvas92 | 3034.0 | 3 | 947 | 347 | 2021-06-28T18:12:51.853Z |
| 4 | 372961 | sunking73 | 2995.5 | 4 | 599 | 251 | 2021-06-20T11:44:49.214Z |
| 5 | 428341 | NPiet | 2995.0 | 5 | 979 | 193 | 2021-06-25T07:22:35.945Z |
| 6 | 4410 | wurstfriedl | 2987.0 | 6 | 2208 | 932 | 2021-06-28T14:49:11.031Z |
| 7 | 3831 | alexttbcn | 2980.0 | 7 | 1303 | 618 | 2021-06-27T18:11:37.414Z |
| 8 | 11002 | twISTer1979 | 2978.6 | 8 | 2411 | 931 | 2021-06-28T17:13:18.551Z |
| 9 | 20585 | fmwallin | 2960.0 | 9 | 1553 | 931 | 2021-06-15T08:53:41.094Z |
[14]:
lb_elo = [f.get_elo_history().rename(columns={"elo":f.name}) for f in
eleven.get_leaderboard() if f.get_elo_history() is not None]
lb_elo.append(p.get_elo_history().rename(columns={"elo":p.name}))
lb_elo_df = reduce(lambda df1,df2: pd.merge(df1,df2,how='outer', left_index=True, right_index=True),
lb_elo).ffill(axis = 0)
[15]:
lb_elo_df.plot(figsize=(10,8))
[15]:
<matplotlib.axes._subplots.AxesSubplot at 0x7fcc34838100>
The dynamics are quite different compared to the one from my cohort of friends. At the top of the leaderboard, we observe rapid Elo growth across the board. The Top 2 player (PROTT_VR), for instance, moved from a 1500 Elo to an impressive 3000+ Elo in a very short period of time. The number 1 player (yedat) has recently surpassed PROTT_VR also showing an amazing Elo growth. I have added my Elo evolution for comparison (the little blue curve at the bottom) to give a perspective on how competitive the game is at the highest level.