错误-字符串索引必须是整数； 使用pandas列作为字典的值

Question

I am practicing coding on dataquest.io, where they provide this code as a way to do random sampling.

wnba['Pts_per_game'] = wnba['PTS'] / wnba['Games Played']

# Stratifying the data in five strata
stratum_G = wnba[wnba.Pos == 'G']
stratum_F = wnba[wnba.Pos == 'F']
stratum_C = wnba[wnba.Pos == 'C']
stratum_GF = wnba[wnba.Pos == 'G/F']
stratum_FC = wnba[wnba.Pos == 'F/C']

points_per_position = {}
for stratum, position in [(stratum_G, 'G'), (stratum_F, 'F'), (stratum_C, 'C'),
                (stratum_GF, 'G/F'), (stratum_FC, 'F/C')]:

    sample = stratum['Pts_per_game'].sample(10, random_state = 0) # simple random sapling on each stratum
    points_per_position[position] = sample.mean()

position_most_points = max(points_per_position, key = points_per_position.get)

I have tried to modify it by grouping the strata into a dictionary, in the following way.

wnba['Pts_per_game'] = wnba['PTS']/wnba['Games Played']

strata = {'stratum_F': wnba[wnba.Pos == 'F'],
'stratum_G': wnba[wnba.Pos == 'G'] , 
'stratum_C': wnba[wnba.Pos == 'C'] ,
'stratum_GF': wnba[wnba.Pos == 'G/F'] ,
'stratum_FC': wnba[wnba.Pos == 'F/C'] }

points_per_position = {}
for stratum, position in strata.items():
    sample = stratum['Pts_per_game'].sample(10,random_state=0)
    points_per_position[position]=sample.mean()

position_most_points=max(points_per_position,key= points_per_position.get)

However, I get TypeError: string indices must be integers . I have tried to work around the stratum['Pts_per_game'] part but could not find the problem.

Answer 1

for stratum, position in strata.items(): means stratum will be a key and position will be the value. the keys in your dict are strings, so stratum is a string

try this:

wnba['Pts_per_game'] = wnba['PTS']/wnba['Games Played']

positions = ['F', 'G', 'C', 'G/F', 'F/C']

strata = {position: wnba[wnba.Pos == position] for position in positions}

points_per_position = {}
for position, stratum in strata.items():
    sample = stratum['Pts_per_game'].sample(10,random_state=0)
    points_per_position[position]=sample.mean()

position_most_points=max(points_per_position,key= points_per_position.get)

What I changed:

• the keys to the dict are now the positions themselves:
• when iterating .items() I unpack the position first and stratum second