Import libraries and dataset.¶

In [1]:
import pandas as pd

elo_history = pd.read_csv("C:\\Users\\chris\\Documents\\Datasets\\FFL\\elo_history.csv")

Get list of teams.¶

In [21]:
team_list = list(elo_history.team_1.unique())
team_list
Out[21]:
['CHRIS',
 'ANDREW',
 'STEVE',
 'MATT',
 'KYLE',
 'CARLEN',
 'NATHAN',
 'DREW',
 'SAM',
 'DYLAN',
 'NICK',
 'JASON',
 'NOAH']

Create "Dates" column.¶

In [27]:
dates = list(elo_history.date.unique())
dates.insert(0, "2014-00")

Initialize team elo rating columns at 0.¶

In [28]:
elo_dict = {}

for i in team_list:
    elo_dict[i] = [0] * len(dates)

Populate columns with data from elo history dataset.¶

In [29]:
curr_date = ""
date_index = 0
for index,row in elo_history.iterrows():
    if row["date"] != curr_date:
        curr_date = row["date"]
        date_index += 1
        
    elo_dict[row["team_1"]][date_index] = round(row["team1_elo_post"])
    elo_dict[row["team_2"]][date_index] = round(row["team2_elo_post"])

# Add initial rating value of 1500 prior to team's first result.
for i in team_list:
    for index,j in enumerate(elo_dict[i]):
        if (j != 0):
            elo_dict[i][index - 1] = 1500
            break

# Replace values of 0 (due to playoff byes) with team's current rating value.
for i in team_list:
    for index,j in enumerate(elo_dict[i]):
        if (j == 0):
            if (index > 0) & (index < len(elo_dict[i]) - 1):
                if (elo_dict[i][index - 1] != 0) & (elo_dict[i][index + 1] != 0):
                    elo_dict[i][index] = elo_dict[i][index - 1]
In [15]:
elo_history
Out[15]:
date team_1 team_2 win team1_elo_pre team2_elo_pre team1_elo_post team2_elo_post
0 2014-01 CHRIS CARLEN 0 1500.000000 1500.000000 1485.000000 1515.000000
1 2014-01 ANDREW KYLE 0 1500.000000 1500.000000 1485.000000 1515.000000
2 2014-01 STEVE DREW 1 1500.000000 1500.000000 1515.000000 1485.000000
3 2014-01 MATT NATHAN 1 1500.000000 1500.000000 1515.000000 1485.000000
4 2014-02 MATT CHRIS 1 1515.000000 1485.000000 1528.708005 1471.291995
... ... ... ... ... ... ... ... ...
559 2021-15 ANDREW STEVE 1 1566.736021 1544.313236 1580.769293 1530.279964
560 2021-15 NICK CHRIS 0 1524.258537 1642.915931 1514.191088 1652.983380
561 2021-15 JASON SAM 0 1483.865477 1506.451432 1469.839220 1520.477689
562 2021-15 DYLAN KYLE 0 1440.436529 1486.708300 1427.422514 1499.722315
563 2021-15 MATT NOAH 1 1368.148791 1492.111812 1388.284599 1471.976003

564 rows × 8 columns

Merge team elo rating columns and date column.¶

In [31]:
df = pd.DataFrame({"DATE" : dates} | elo_dict)
df.tail()
Out[31]:
DATE CHRIS ANDREW STEVE MATT KYLE CARLEN NATHAN DREW SAM DYLAN NICK JASON NOAH
112 2021-11 1644 1521 1523 1369 1533 0 0 0 1487 1454 1518 1465 1542
113 2021-12 1654 1533 1513 1390 1511 0 0 0 1505 1442 1530 1453 1524
114 2021-13 1663 1547 1529 1381 1502 0 0 0 1491 1427 1540 1467 1509
115 2021-14 1643 1567 1544 1368 1487 0 0 0 1506 1440 1524 1484 1492
116 2021-15 1653 1581 1530 1388 1500 0 0 0 1520 1427 1514 1470 1472

Save and export table to CSV.¶

In [32]:
df.to_csv("C:\\Users\\chris\\Desktop\\elo_over_time.csv", index=False)