I try to insert np.nan values into an array at the nth position. Desired output:
array(['1', 'nan', 'nan', '2', test', '3'])
I tried with this code:
position = [1,2]
array= np.array([1,2,'test', 3])
array= np.insert(array, position, np.nan)
But it is inserting the values at index 1 and 3:
array(['1', 'nan', '2', 'nan', 'test', '3'])
How can I fix this?
Inserting would take place altogether for both np.nan. So, if you use [1,2] then index 1 would be between 1 and 2, and index 2 would be between 2 and 'test'.
The position needs to be [1,1] if you want to insert continuous values.
import numpy as np
position = [1,1]
array = np.array([1,2,'test', 3])
array = np.insert(array, position, np.nan)
print(array)
Output:
['1' 'nan' 'nan' '2' 'test' '3']
The position should be [1, 1]. Position 2 is between 2 and 'test' and 1 is between 1 and 2. The index where you insert them is where the index is located in the initial array, not where they will end up.
The numpy.insert() function inserts values before the given indices.
i.e. for the given array :
[1, 2,'test',3]
0 1 2 3 -> index
Inserting at position=[1,2] will insert values before index 1 (value =2) and before index 2(value ='test') in the original array i.e.
[1, ___, 2, ___, 'test',3]
In order to get both nan inserted before index 1, write as
import numpy as np
position = [1,1]
array= np.array([1,2,'test', 3])
array= np.insert(array, position, np.nan)
print(array)
You will get the desired result:
['1' 'nan' 'nan' '2' 'test' '3']
Note: This position array is not the index of the new values you want to insert, it is the index before which you want to insert the new values
arraydefinition has a mix of numbers and str, the result is string dtype. And thenp.nanis inserted as a string. In a comment you talk about replacing 'string types' withnanin adataset. That sounds more like apandasoperation. Remembernp.nanis afloat. Do you want to mix integers, floats, and strings?