Odtnhj

I have this loop that continuously checks api data and add it to an array.
Then I calculate an indicator based on one column of the array that I want to add the the array. The issue only is that the indicator need 10 data point to get going.

import numpy as np

list_b = 
i = 0

while True:
 i += 1
 list_a = 

 list_a.append(i)
 list_a.append(33)
 list_a.append(44)
 list_b.append(list_a)
 array = np.array(list_b, dtype=float)

 if len(array) > 10:
 x = indicator_function(array[:,0])
 np.append(x) # in a 4th column

 if i == 15:
 break

print(array.shape)
print(len(array))
print(array)

The issue is that I cant append to the array because it doesnt have the same dimensions as the array before. So how can I extend the array by one column or insert the new indicators values in a new column after 10 rows.

The array should be formed like this:

1 33 44
2 33 44
...
9 33 44
10 33 44
11 33 44 x
12 33 44 x

From what I understand, numpy doesn't allow for jagged arrays (arrays with a changing number of columns). You might have to create an entirely new array. Or, if you want to do something crazy, you could move the if statement up above list_b.append(list_a) and change it to

if (i > 10):
 x = indicator_function(array[:,0])
 list_a.append(x)
else:
 list_a.append(None)

if i < 10: 
 list_a.append(i)
 list_a.append(33)
 list_a.append(44)
 list_a.append(55)
 list_b.append(list_a)
 array = np.array(list_b, dtype=float)

if i > 10:
 list_a.append(i)
 list_a.append(33)
 list_a.append(44)
 list_a.append(66)
 list_b.append(list_a)
 array = np.array(list_b, dtype=float)

@sahil yes that works.

There's no builtin support for proper jagged arrays in numpy. The best bet for your application is probably a masked array:

import itertools
import numpy.ma as ma

array = None
list_b = 

def indicator_function(arr):
 return -1

for i in itertools.count():
 list_a = 
 list_b.append(list_a)

 list_a.append(i)
 list_a.append(33)
 list_a.append(44)
 list_a.append(np.nan)
 array = ma.MaskedArray(list_b)

 # assign mask or indicator_function, as appropriate
 ix = 10 + 1
 array[:ix, -1] = ma.masked
 array[ix:, -1] = indicator_function(array[:,0])

 if i == 15:
 break

This results in a masked array that looks like:

[[0.0, 33.0, 44.0, --],
 [1.0, 33.0, 44.0, --],
 [2.0, 33.0, 44.0, --],
 [3.0, 33.0, 44.0, --],
 [4.0, 33.0, 44.0, --],
 [5.0, 33.0, 44.0, --],
 [6.0, 33.0, 44.0, --],
 [7.0, 33.0, 44.0, --],
 [8.0, 33.0, 44.0, --],
 [9.0, 33.0, 44.0, --],
 [10.0, 33.0, 44.0, --],
 [11.0, 33.0, 44.0, -1.0],
 [12.0, 33.0, 44.0, -1.0],
 [13.0, 33.0, 44.0, -1.0],
 [14.0, 33.0, 44.0, -1.0],
 [15.0, 33.0, 44.0, -1.0]]