I wanted to create a Python dictionary from 2 DataFrame columns. I had this DF which may be familiar to Scottish Football fans:
Position Team GP W D L GF GA GD Pts PPG
0 1 Celtic 38 27 6 5 77 20 57 87 2.29
1 2 Rangers 38 23 9 6 82 27 55 78 2.05
2 3 Kilmarnock 38 19 10 9 50 31 19 67 1.76
3 4 Aberdeen 38 20 7 11 57 44 13 67 1.76
4 5 Hibernian 38 14 12 12 51 39 12 54 1.42
5 6 St Johnstone 38 15 7 16 38 48 -10 52 1.37
6 7 Hearts 38 15 6 17 42 50 -8 51 1.34
7 8 Motherwell 38 15 6 17 46 56 -10 51 1.34
8 9 Livingston 38 11 11 16 42 44 -2 44 1.16
9 10 Hamilton 38 9 6 23 28 75 -47 33 0.87
10 11 St Mirren 38 8 8 22 34 66 -32 32 0.84
11 12 Dundee FC 38 5 6 27 31 78 -47 21 0.55
I specifically wanted a dictionary with entries of Team and Position. It only took a few minutes to find this code stackoverflow:
NewDF = pd.Series(df.Team,index=df.Position).to_dict()
Which Gives the result:
{1: 'Rangers',
2: 'Kilmarnock',
3: 'Aberdeen',
4: 'Hibernian',
5: 'St Johnstone',
6: 'Hearts',
7: 'Motherwell',
8: 'Livingston',
9: 'Hamilton',
10: 'St Mirren',
11: 'Dundee FC',
12: nan}
Of course, this isn’t right. Sevco will never be No. 1!!
It seems that Position1 doesn’t show the correct value. I grew tired of trying to fix this after an hour or so.
The best way I found was:
b = dict(zip(a.Team, a.Position))
This is simply using the Python zip method on the 2 columns and forming a dictionary from them.
Zip takes 2 iterables and makes tuples ( a,b). The dict function changes them into s dictionary type.
It gives this result:
{'Celtic': 1,
'Rangers': 2,
'Kilmarnock': 3,
'Aberdeen': 4,
'Hibernian': 5,
'St Johnstone': 6,
'Hearts': 7,
'Motherwell': 8,
'Livingston': 9,
'Hamilton': 10,
'St Mirren': 11,
'Dundee FC': 12}
The world is put to rights!! But why didn’t it work on my original DF?
if I make a DF from scratch:
xx =[1,2,3,4,5 ]
yy = ['a','b','c','d','e']
d = {'Month':xx,'Text':yy}
df = pd.DataFrame(d)
month Text 0 1 a 1 2 b 2 3 c 3 4 d 4 5 e
However the following code gives me the same problem as my original;
zz = pd.Series(df.Text,index=dx.Month).to_dict()
{1: ‘b’, 2: ‘c’, 3: ‘d’, 4: ‘e’, 5: nan}
This gives the same problem as my original one. However:
aa = dict(zip(df.Text, df.month))
gives the correct result (or at least the ones I want)
{‘a’: 1, ‘b’: 2, ‘c’: 3, ‘d’: 4, ‘e’: 5}
The pd.series version was found on stackoverflow. It doesn’t work for me and I won’t spend time trying to figure out why.
Conclusion
If you want to make a dictionary from 2 DF columns, then use the inbuilt zip method and create a dictionary.
