Finding connected components in a pixel-array - python

I have a pixel-array like the array below and from that I want to distinguish the two "groups" of 1s. The plan is to do this in a large set of similar pixel-arrays so I need to find a way to do this efficient.
Maybe I can add all the 1-positions to a separate array and do some search to find the ones connected, but it should be some better way. Is there any algorithms for finding connected components like this?
[
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
[0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
[0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0]
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0]
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
[0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
[0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
[0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
[0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
[0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
[0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
[0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
[0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
[0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
]


Considering that the groups should never touch each other, you can use scipy.ndimage.measurements.label to label the groups:
In [1]: import numpy as np
In [2]: from scipy.ndimage.measurements import label
In [3]: array = np.array(...) # your example
In [4]: structure = np.ones((3, 3), dtype=np.int) # this defines the connection filter
In [5]: structure # in this case we allow any kind of connection
Out[5]:
array([[1, 1, 1],
[1, 1, 1],
[1, 1, 1]])
In [6]: labeled, ncomponents = label(array, structure)
In [7]: labeled
Out[7]:
array([[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 2, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 2, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 2, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 2, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]], dtype=int32)
In [7]: ncomponents
Out[7]: 2
Although I haven't read the particular implementation, SciPy tends to use highly efficient algorithms implemented in C, hence the performance should be relatively high. You can then extract the indices for each group using NumPy:
In [8]: indices = np.indices(array.shape).T[:,:,[1, 0]]
In [9]: indices[labeled == 1]
Out[9]:
array([[ 1, 6],
[ 1, 7],
[ 2, 6],
[ 2, 7],
[ 2, 8],
[ 2, 9],
[ 2, 10],
[ 2, 11],
[ 2, 12],
[ 2, 13],
[ 3, 11],
[ 3, 12],
[ 3, 13]])
In [10]: indices[labeled == 2]
Out[10]:
array([[ 5, 1],
[ 6, 1],
[ 7, 1],
[ 7, 2],
[ 8, 1],
[ 8, 2],
[ 9, 2],
[10, 2],
[10, 3],
[11, 2],
[11, 3],
[12, 3],
[13, 3]])

Related

python cv2.findContours does a bad job finding contours in a simple grayscale image

I have the following back and white image
import numpy as np
thresh = np.array([[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0],
[0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0],
[0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0],
[0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1],
[0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1],
[0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1],
[0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1],
[0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1],
[0, 0, 0, 0, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0, 1, 1, 1, 1]]).astype('uint8')
I am trying to find contour in the thresh image, like this
import cv2
contours, hierarchy = cv2.findContours(
thresh,
cv2.RETR_CCOMP,
cv2.CHAIN_APPROX_SIMPLE
)
Just looking at the threshold image, it's intuitive that there is 1 big contour around the 1, that is shaped like an arrow.
However, visually inspection of the returned contour from cv2 found using
canvas = np.zeros_like(thresh)
for ct in contours:
cv2.drawContours(canvas, ct,-1, 1, 1)
yeilds the following;
array([[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1],
[0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 1, 0, 0, 1]],
dtype=uint8)
In summary, how would be best get the contour from thresh image?

The result you obtained is correct.
Contours allow you to find the points along the boundary of any shape. The third parameter in cv2.findContours lets you decide how you want to store the boundary points. You have 2 ways of doing that (1) either store all the points OR (2) find a good approximation.
In your case, you are using the flag cv2.CHAIN_APPROX_SIMPLE. This method does not store all the boundary points of the shape. For every line along the boundary, it stores just 2 points (the ends of each line). This is the best way to approximate the shape of any contour and it also saves memory.
If you want to store all the boundary points you need to use cv2.CHAIN_APPROX_NONE.
Here's the documentation for more
Illustration:
Consider the following array as input:
thresh = np.array([[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]],
dtype=uint8)
Using the flag cv2.CHAIN_APPROX_SIMPLE you would get:
array([[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]],
dtype=uint8)
And using the flag cv2.CHAIN_APPROX_NONE you would get:
array([[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]],
dtype=uint8)

I have a python list of lists that was stored as a string and need to reverse it [duplicate]

This question already has answers here:
How to convert string representation of list to a list
(19 answers)
Closed 1 year ago.
I had a list of lists (in fact a small image) that unfortunately someone had converted it to a string before it was stored. Like this:
'[[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]]'
I am trying to find a way to reverse it to the obvious list format it represents without having to build a parser myself. Unfortunately I have not found any answers of how to do that> it probably is simple and obvious but since searching for list and string returns a myriad of other things I have been facing some difficulties.

You can use eval to do this, but the safer alternative is ast.literal_eval.
import ast
s = '[[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]]'
l = ast.literal_eval(s)

Train test split a list of sentence

I have a list of sentences. I want to randomly separate into 80% and 20%, it looks like this:
['Hi.',
'Hi.',
'Run!',
'Wow!',
'Wow!',
'Fire!',
'Help!',
'Help!',
'Stop!',
'Wait!',
'Go on.',
'Hello!',
'I ran.',
'I see.',
'I see.',
'I try.',
'I won!',...]
I was thinking using a mask
import random
mask = [0] * 4000 + [1] * 16000
random.shuffle(mask)
But it is not like a data frame.
and I tried
percent=80
bol_mask =[random.randrange(100) < percent for i in range(100)]
Cant really apply boolean to sentences
Also the separation mask must be kept, and will later apply to another list in German, which is the corresponding translation.
it looks like this
array([[ 553, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0],
[3430, 1114, 6, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0],
[1115, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0],
[3431, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0],
[3432, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0],
[2459, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0],
[3433, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0],
[1533, 3434, 6, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0],
[2460, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 394, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0]],
dtype=int32)
My question is how to apply mask to a list of sentences? and keep the same split and apply to the corresponding ndarray?

Actually I've solved it my self.
bol_mask =[random.randrange(100) < 80 for i in range(20000)]
inv_mask = np.invert(bol_mask)
Eng_train =np.array(Eng)[bol_mask]
Eng_test =np.array(Eng)[inv_mask]
German_train = padded[bol_mask]
German_test = padded[inv_mask]
Thanks Grayrigel, an accept for your effort in helping

If using scikit-learn is an option, you can just use train_test_split method as the following:
>>> from sklearn.model_selection import train_test_split
>>> print(x)
>>> x
['Hi.', 'Hi.', 'Run!', 'Wow!', 'Wow!', 'Fire!', 'Help!', 'Help!', 'Stop!', 'Wait!']
>>> len(x)
10
>>> x1
array([[ 553, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0],
[3430, 1114, 6, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0],
[1115, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0],
[3431, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0],
[3432, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0],
[2459, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0],
[3433, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0],
[1533, 3434, 6, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0],
[2460, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0],
[ 394, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0]])
>>> x1.shape
(10, 20)
#assuming x, x1 have same length train test split should work fine.
>>> train, test, train_german, test_german = train_test_split(x,x1, test_size=0.2, shuffle=True)
>>> len(train)
8
>>> len(test)
2
>>> len(train_german)
8
>>> len(test)
2

How to solve 4 x 4 array with 16 unknown with Python (Update)

I need to solve a 2 by 2 array with 4 unknown
A B
C D
I know all horizontal sum A+B=11, C+D=7
I know all vertical sum A+C=10, B+D=8
I know all diagonal sum A+D=15, B+C=3
I then use Python to solve for A,B,C,D
import numpy as np
A = [[1, 1, 1, 1],
[1, 0, 0, 1],
[1, 0, 1, 0],
[0, 0, 1, 1]]
a = [18, 15, 10, 7]
answera = np.linalg.solve(A, a)
print(answera)
And the answer is [9. 2. 1. 6.] which is correct
Now I need to solve 4 by 4 array with 16 unknown
A B C D
E F G H
I J K L
M N O P
I know horizontal sum A+B+C+D=10, E+F+G+H=26, I+J+K+L=42, M+N+O+P=58
I know vertical sum A+E+I+M=28, B+F+J+N=32, C+G+K+O=36, D+H+L+P=40
I know diagonal sum M=13, I+N=23, E+J+O=30, A+F+K+P=34, B+G+L=21, C+H=11, D=4
The other diagonal sum A=1, B+E=7, C+F+I=18, D+G+J+M=34, H+K+N=33, L+O=27, P=16
Which mean I know the value of the 4 corners.
I tried the following code but did not work
C = [[1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1],
[0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0],
[0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0],
[1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1],
[0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0],
[0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0],
[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]]
c = [10, 26, 42, 58, 7, 21, 39, 33, 27, 11, 23, 35, 30, 23, 32, 136]
answerc = np.linalg.solve(C, c)
print(answerc)
The correct answer should be [1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.] but I got error message
Traceback (most recent call last):
answerc = np.linalg.solve(C, c)
r = gufunc(a, b, signature=signature, extobj=extobj)
raise LinAlgError("Singular matrix")
numpy.linalg.LinAlgError: Singular matrix
Am I in the right direction? I will need to solve 5X5 with 25 unknown, 6X6 with 36 unknown and so on. Is there an easier way?
-----------------------------------------------------------------------------
Following Mr. Rory Daulton solution, I can solve the above 1 to 16 4X4 array without problem, but when I use it in another array with negative number, it doesn't give answer as expected;
The negative 4X4 array as follow
-20 -10 -5 0
-10 -20 -10 -5
-5 0 -10 -20
-10 -20 -10 -5
My python code as follow
import numpy as np
G = [[1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # horizontal rows
[0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1],
[1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0], # vertical columns
[0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0],
[0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0],
[0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0], # forward diagonals
[0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0],
[0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0],
[1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1],
[0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0],
[0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # back diagonals
[0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1],
]
g = [-35, -45, -35, -45, # horizontal rows
-45, -50, -35, -30, # vertical columns
-10, -25, -20, -55, -40, -10, 0, # forward diagonals
-20, -20, -30, -20, -35, -30, -5, # back diagonals
]
answerg = np.linalg.lstsq(G, g, rcond=None)
print(answerg[0])
The output is not exactly the original array
[-2.00000000e+01 -1.31250000e+01 -1.87500000e+00 8.88178420e-15
-6.87500000e+00 -2.00000000e+01 -1.00000000e+01 -8.12500000e+00
-8.12500000e+00 2.13162821e-14 -1.00000000e+01 -1.68750000e+01
-1.00000000e+01 -1.68750000e+01 -1.31250000e+01 -5.00000000e+00]
What should I try? Thank you in advance.

SHORT ANSWER: There are infinitely many solutions to your problem. So this takes a more complex analysis of the equations.
LONG ANSWER: You have multiple problems with your code.
First, you make it easy to make mistakes, since the lines of your matrix do not correspond to the data that you present. Worse, you have no comments to explain things. This mis-match will probably cause errors. You have 22 pieces of data in your sums, so use them. You tried to combine some of the sums and ignore others (the four corners) but you did not do it properly and you ended up with a singular matrix.
Next, you use linalg.solve. In your problem you have more data items (22) than unknowns (16), so solve is inappropriate. The numpy documentation for solve states
a must be square and of full-rank, i.e., all rows (or, equivalently,
columns) must be linearly independent; if either is not true, use
lstsq for the least-squares best “solution” of the system/equation.
The matrix resulting from your data is not square, therefore the rows are not linearly independent, so you should use lstsq rather than solve. The lstsq routine gives more information than you need for your problem, so just print the first item in the resulting list.
Combining those ideas and adding a few comments gives this code:
import numpy as np
C = [[1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # horizontal rows
[0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1],
[1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0], # vertical columns
[0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0],
[0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0],
[0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0], # forward diagonals
[0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0],
[0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0],
[1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1],
[0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0],
[0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # back diagonals
[0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1],
]
c = [10, 26, 42, 58, # horizontal rows
28, 32, 36, 40, # vertical columns
13, 23, 30, 34, 21, 11, 4, # forward diagonals
1, 7, 18, 34, 33, 27, 16, # back diagonals
]
answerc = np.linalg.lstsq(C, c, rcond=None)
print(answerc[0])
The printout is what you want:
[ 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.]
However, to be honest, there is no guarantee that this is an answer--just that it is a "closest" answer. Also, if it is an answer, there may be other answers. And, in fact, further analysis shows that there are other answers that satisfy all your conditions.
The sympy module can generate a row reduced echelon form of the matrix, which can be used to do more in-depth analysis of all the answers. However, the constants are then to be part of the matrix, rather than used as a separate array. Here is code for sympy to attempt to solve your problem:
import sympy
C = [[1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 10], # horizontal rows
[0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 26],
[0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 42],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 58],
[1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 28], # vertical columns
[0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 32],
[0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 36],
[0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 40],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 13], # forward diagonals
[0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 23],
[0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 30],
[1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 34],
[0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 21],
[0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 11],
[0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 4],
[1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1], # back diagonals
[0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7],
[0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 18],
[0, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 0, 34],
[0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 33],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 27],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 16],
]
print(sympy.Matrix(C).rref())
The printout is
(Matrix([
[1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1],
[0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, -1, 0, -13],
[0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 18],
[0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 4],
[0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 20],
[0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 6],
[0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7],
[0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, -1, 0, -7],
[0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, -1, 0, -6],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 10],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 11],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 27],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 13],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 29],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 16],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]]), (0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 15))
If you know how to analyze this you will see that there are infinitely many answers to your problem. If you set the number in the 15th cell to 15+x then the above matrix shows that the answer to all your restrictions is
1 2+x 3-x 4
5-x 6 7 8+x
9+x 10 11 12-x
13 14-x 15+x 16
The solve function of numpy only works if there is just one solution, so even if you had adjusted your matrix differently it would not have worked for you.
ANSWER TO YOUR UPDATE:
It seems that you missed the point of my answer. Your 4x4 problem has infinitely many answers, so there is no procedure that can choose the particular answer that you have in mind. The np.linalg.lstsq routine can find one of the answers to your problem but probably will not find your desired answer. You should consider it to be a coincidence that using that routine in your first problem gave your desired answer--that will probably not work in other problems.
It is a little hard to interpret the given answer to your new problem, since the scientific notation is hard to read. But all those matrix values are exact, and here they are as rational numbers in a particular format that should be obvious:
-20 -10-(3+1/8) - 5+(3+1/8) 0
-10+(3+1/8) -20 -10 - 5-(3+1/8)
- 5-(3+1/8) 0 -10 -20+(3+1/8)
-10 -20+(3+1/8) -10-(3+1/8) - 5
You see that the numpy's answer is the one that you expected, with the value 3+1/8 added to or subtracted from half the array values. This makes x=3+1/8 in the general answer that I gave you for your first problem.
This is as good as you can expect. Numpy gave you a correct answer--it has no idea how to choose the answer that you had in your head from the infinitely many correct answers to your problem. The only way to get just one answer is to change your problem--perhaps state the value in the first row and second column, or the sum of the first and third values in any one of the rows, or something similar.

How should we pad text sequence in keras using pad_sequences?

I have coded a sequence to sequence learning LSTM in keras myself using the knowledge gained from the web tutorials and my own intuitions. I converted my sample text to sequences and then padded using pad_sequence function in keras.
from keras.preprocessing.text import Tokenizer,base_filter
from keras.preprocessing.sequence import pad_sequences
def shift(seq, n):
n = n % len(seq)
return seq[n:] + seq[:n]
txt="abcdefghijklmn"*100
tk = Tokenizer(nb_words=2000, filters=base_filter(), lower=True, split=" ")
tk.fit_on_texts(txt)
x = tk.texts_to_sequences(txt)
#shifing to left
y = shift(x,1)
#padding sequence
max_len = 100
max_features=len(tk.word_counts)
X = pad_sequences(x, maxlen=max_len)
Y = pad_sequences(y, maxlen=max_len)
After a carefully inspection I found my padded sequence looks like this
>>> X[0:6]
array([[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 5],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 4],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7]], dtype=int32)
>>> X
array([[ 0, 0, 0, ..., 0, 0, 1],
[ 0, 0, 0, ..., 0, 0, 3],
[ 0, 0, 0, ..., 0, 0, 2],
...,
[ 0, 0, 0, ..., 0, 0, 13],
[ 0, 0, 0, ..., 0, 0, 12],
[ 0, 0, 0, ..., 0, 0, 14]], dtype=int32)
Is the padded sequence suppose to look like this? Except the last column in the array the rest are all zeros. I think I made some mistake in padding the text to sequence and if so can you tell me where I made the error?

If you want to tokenize by char, you can do it manually, it's not too complex:
First build a vocabulary for your characters:
txt="abcdefghijklmn"*100
vocab_char = {k: (v+1) for k, v in zip(set(txt), range(len(set(txt))))}
vocab_char['<PAD>'] = 0
This will associate a distinct number for every character in your txt. The character with index 0 should be preserved for the padding.
Having the reverse vocabulary will be usefull to decode the output.
rvocab = {v: k for k, v in vocab.items()}
Once you have this, you can first split your text into sequences, say you want to have sequences of length seq_len = 13 :
[[vocab_char[char] for char in txt[i:(i+seq_len)]] for i in range(0,len(txt),seq_len)]
your output will look like :
[[9, 12, 6, 10, 8, 7, 2, 1, 5, 13, 11, 4, 3],
[14, 9, 12, 6, 10, 8, 7, 2, 1, 5, 13, 11, 4],
...,
[2, 1, 5, 13, 11, 4, 3, 14, 9, 12, 6, 10, 8],
[7, 2, 1, 5, 13, 11, 4, 3, 14]]
Note that the last sequence doesn't have the same length, you can discard it or pad your sequence to max_len = 13, it will add 0's to it.
You can build your targets Y the same way, by shifting everything by 1. :-)
I hope this helps.

The problem is in this line:
tk = Tokenizer(nb_words=2000, filters=base_filter(), lower=True, split=" ")
When you set such split (by " "), due to nature of your data, you'll get each sequence consisting of a single word. That's why your padded sequences have only one non-zero element. To change that try:
txt="a b c d e f g h i j k l m n "*100

The argument padding controls padding either before or after each sequence. Use like this:
X = pad_sequences(x, maxlen=max_len, padding='post')
Y = pad_sequences(y, maxlen=max_len, padding='post')

Categories