CODEWITHSUNDEEP
pythonpandas
user466534
user466534

Reputation:

mapping list to the binary data

let us suppose that in my dataframe, we have following categorical data [3,4,5,6,7,8,9], i want to map values [3,4,5,6,7] to the zero and [8,9] to the one, of course i can use np.where function, but i have tried dictionary method like this

winequality["quality"] =winequality["quality"].map({[3, 4, 5, 6, 7]:0,[8, 9]:1})

where winequality is dataframe and quality is column, what i have got is this error

TypeError: unhashable type: 'list'

i have tried such kind of mapping and it was working fine, so exactly i dont remember which part i am missing, could you help me please?

Upvotes: 0

Views: 290

Answers (3)

Shubham Sharma
Shubham Sharma

Reputation: 71707

Setup

>>> winequality

   quality
0        3
1        4
2        5
3        6
4        7
5        7
6        8
7        9
8        9
9        8

1. Series.map

Instead of list we can create a dictionary with tuple as its keys, then flatten the dictionary, and map the values

dct = {(3, 4, 5, 6, 7):0, (8, 9): 1}
dct = {i:v for k, v in dct.items() for i in k}

winequality['quality'].map(dct)

2. Series.replace

replace can be used without requiring to flatten the dictionary first.

dct = {(3, 4, 5, 6, 7):0, (8, 9): 1}
winequality['quality'].replace(dct)

Result

0    0
1    0
2    0
3    0
4    0
5    0
6    1
7    1
8    1
9    1
Name: quality, dtype: int64

Upvotes: 4

divingTobi
divingTobi

Reputation: 2300

I believe the part where you try to create the dictionary is wrong: {[3, 4, 5, 6, 7]:0,[8, 9]:1}. You should replace this with

d1 = {k: 0 for k in [3,4,5,6,7]}
d2 = {k: 1 for k in [8,9]}
d = {**d1, **d2}

That would give you

{1: 0, 3: 0, 4: 0, 9: 1, 10: 1}

for d, which you can then use in .map.

Upvotes: 2

tdy
tdy

Reputation: 41437

If you want to map() it, the dict keys need to be enumerated, not as a list, e.g.:

zeros = {key: 0 for key in [3, 4, 5, 6, 7]}
ones = {key: 1 for key in [8, 9]}
winequality["quality"] = winequality["quality"].map({**zeros, **ones})

Upvotes: 3

Related Questions