How to fill a figure with a certain number of subplots?

Try replacing this:

rows = int(round(subplots/20))

with this:

rows = int(round(subplots/cols))

That will fix the the calculated amount of subplots.

Since plt.subplots() define a rectangular grid of subplots, the total amount of them should not be a prime number, because it must be divisible by the number of columns and the number of rows. If you enter a prime number as subplots, you code will find the nearest not-prime number. For example, if you enter:

subplots = 149

your code will give you 150 subplots, because it does not exist a combination of cols * rows which gives 149 as a product.

In order to manage this, use this code:

import matplotlib.pyplot as plt

subplots = 149
cols = int(round(subplots/10))
rows = int(round(subplots/cols))

print(f'subplots: {cols*rows}')

if subplots > cols * rows:
    fig, axes = plt.subplots(nrows = rows + 1, ncols = cols)
else:
    fig, axes = plt.subplots(nrows = rows, ncols = cols)

_axes = []
for ax_array in axes:
    for ax in ax_array:
        _axes.append(ax)

for ax in _axes:
    ax.set_xticklabels([])
    ax.set_yticklabels([])
    ax.tick_params(axis='both', which='both', length=0)
    ax.plot()

if subplots > cols * rows:
    for idx in range(1, cols):
        fig.delaxes(axes[rows, idx])
elif subplots < cols * rows:
    fig.delaxes(axes[rows - 1, cols - 1])

plt.tight_layout()
plt.subplots_adjust(wspace=0, hspace=0)
plt.show()

• in the case when subplots > rows * cols (for example subplots = 31) it will add a subplot at the bottom:

• in the case when subplots < rows * cols (for example subplots = 23) it will delete the last subplot:

So you can manage all cases.

I found a workaround using the modulo operator and removing excessive subplots afterwards. In this example 149 subplots are to be depicted in a figure with space for 150 subplots. The figure is expanded to the next number divisible by 10 so in this case to 150.

import matplotlib.pyplot as plt


features = 149
if features % 10 != 0:
    subplots = features + (10-(features % 10))
else:
    subplots = features
print(f'features: {features}')

cols = int(round(subplots/10))
rows = int(round(subplots/cols))
print(f'subplots: {cols*rows}')

fig, axes = plt.subplots(nrows=rows, ncols=cols)
_axes = []
for ax_array in axes:
    for ax in ax_array:
        _axes.append(ax)

# axes to be plotted
for ax in _axes:
    ax.set_xticklabels([])
    ax.set_yticklabels([])
    ax.tick_params(axis='both', which='both', length=0)
    ax.plot()

# axes to be removed
for ax in range(10-(features % 10)):
    fig.delaxes(_axes[-(ax+1)])

plt.tight_layout()
plt.subplots_adjust(wspace=0, hspace=0)
plt.show()

Output:

features: 149
subplots: 150

If you really need to deal with prime numbers and if having a non-squared grid is an option, you could just define a reference square grid and "turn off" the remaining plots:

import matplotlib.pyplot as plt

subplots = 149
cols, rows = 10, 15

nullplots = cols * rows - subplots

print(f'subplots: {subplots}')

fig, axes = plt.subplots(nrows=rows, ncols=cols)

# Supress the nullplots
for idx in range(nullplots):
    axes[14, -(idx + 1)].axis('off')

_axes = []
for ax_array in axes:
    for ax in ax_array:
        _axes.append(ax)

for ax in _axes:
    ax.set_xticklabels([])
    ax.set_yticklabels([])
    ax.tick_params(axis='both', which='both', length=0)
    ax.plot()

plt.tight_layout()
plt.subplots_adjust(wspace=0, hspace=0)
plt.show()

In this slight modification of your code I have defined a 10 x 15 square grid that will host 149 plots. Then I remove the 150 - 149 = 1 remaining plots in the right-bottom corner for example. The resulting plot:

I hope this helps you.