CODEWITHSUNDEEP
pythonpandas
K.lid
K.lid

Reputation: 21

How do I change an index value in a Multi-index Pandas dataframe?

If you look at the above image, the level 0 index (3, 102483) are random numbers. And this goes on until the end of the dataset. I want to be able to change these numbers to be from 0,1,2,3 so on. So 3 would be 0 and so onI have a data frame with a multi index and I wanted to change the values on one of them.

For example:

index = [1,23,356,405,513,65,6787,898,679]
index_2 = ["A","B","C","D","E","F","G","H","I"]
names= ["James","Adam","Mary","Tom","Sam","Harry","Jacob","Isa","Rick"]

df_test = pd.DataFrame(data=names, index=[index, index_2])

This will give me a data frame with two indices. "Index" are random numbers such as the above. However I want to change the index to default values such that it takes values from 0,1,2,3 so on instead of the random numbers above.

I am doing it with a very large dataset which has random numbering to the data and I want to change it to have default index numbers from 0,1,2 and so on.

So my question is, how do I replace the values with a default index?

Upvotes: 2

Views: 1446

Answers (3)

Ben.T
Ben.T

Reputation: 29635

If df_test exist already, you can set the index with pd.MultiIndex.from_arrays and get the codes from the original index level you want to replace by incremental value and get_level_values for the other one.

# assume df_test created like this 
index = [1,1,356,356,356,6787,6787,6787,6787] #change this to be more like your problem
index_2 = ["A","B","C","D","E","A","B","C","D"]
names= ["James","Adam","Mary","Tom","Sam","Harry","Jacob","Isa","Rick"]

df_test = pd.DataFrame(data=names, index=[index, index_2])
print (df_test)
            0
1    A  James
     B   Adam
356  C   Mary
     D    Tom
     E    Sam
6787 A  Harry
     B  Jacob
     C    Isa
     D   Rick

# so you can do to have regular incremental for first level of index
df_test.index = pd.MultiIndex.from_arrays([df_test.index.codes[0], 
                                           df_test.index.get_level_values(1)])
print (df_test)
         0
0 A  James
  B   Adam
1 C   Mary
  D    Tom
  E    Sam
2 A  Harry
  B  Jacob
  C    Isa
  D   Rick

Upvotes: 1

sammywemmy
sammywemmy

Reputation: 28649

not sure if this is what u r after, a visual of ur expected output would be helpful : 

#drop the index with the random numbers
df_test = df_test.droplevel(0)
#get the indices for the letters
#assumption here is that the letters r not repeated
new_index = df_test.index.get_indexer_for(df_test.index)

#if the letters are not unique :
from itertools import chain, islice
c = chain.from_iterable
e = enumerate
#this allows us to pull the unique numbers per letter
new_index = islice(c(e(df_test.index)),0,None,2)

#assign the new index, and set it as the outermost index
df_test.set_index(new_index,append=True).swaplevel(1,0)

Upvotes: 0

Yaakov Bressler
Yaakov Bressler

Reputation: 12008

You could pass a range with the length of the data to your index:

range_1 = list(range(len(names)))
df_test = pd.DataFrame(data=[names], index=[range_1, index_2])

Upvotes: 0

Related Questions