class Algo(Arr):
def __init__(self, win):
super().__init__()
pygame.init()
self.win = win
self.printr = Printr(self.win)
self.draw = Draw(self.win, self.pixels)
self.max_k = int(len(self.arr))
self.k = self.init_k()
## Flags
self.moving = False
self.draw_outer_circle = False
self.survived = True
### Circle ###
self.pixel_mid = [800, 400]
self.pixel_radius = 110
#self.configure_draw()
self.convert(True) ## Initialize self.arr_mid
def configure_draw(self):
self.draw.configure_scales(self.arr_scale, self.pixel_scale)
def update(self, moving, draw_outer):
self.moving = moving
self.draw_outer_circle = draw_outer
self.move()
self.convert(self.moving)
self.knn()
def draw_things(self, win_w, win_h):
self.draw.draw_plot(self.norm_arr)
radius = 0
if self.draw_outer_circle:
radius = self.pixel_radius
self.draw.draw_circle(self.arr_mid, self.pixel_mid, radius,
self.survived)
self.printr.print_instructions(self.k, self.survived, win_w, win_h)
""" UPDATES """
def move(self):
if self.moving:
mx, my = pygame.mouse.get_rel()
## Move circle if the mouse is anywhere on the graph
if pygame.mouse.get_pos()[0] < self.pixel_origin[0] + self.pixel_w:
self.pixel_mid[0] += mx
self.pixel_mid[1] += my
def convert(self, moving):
if moving:
self.norm_arr_mid = self.convert_pixel_to_norm_arr(self.pixel_mid)
self.arr_mid = self.convert_norm_arr_to_arr(self.norm_arr_mid)
def update_k(self, delta_k):
new_k = self.k + delta_k
if new_k > 0 and new_k < self.max_k:
self.k = new_k
""" KNN ENGINE """
def knn(self):
## Reset the distances and nn
self.norm_arr[:, 3:] = 0
## Calculate the distances and nn -- note, need to vectorize
""" Need to vectorize this """
for i, coord in enumerate(self.norm_arr[:, :2]):
euclid = np.sqrt(np.sum(np.square(self.norm_arr_mid - coord)))
self.norm_arr[i:, 3] = euclid
## partition the array into smallest / largest K
part = np.argpartition(self.norm_arr[:, 3], self.k)
self.norm_arr[part[:self.k], 4] = 1
self.update_radius(self.norm_arr[part[:self.k]])
self.update_survived(self.norm_arr[part[:self.k]])
def update_survived(self, part):
self.survived = True
if part[:, 2].sum() < (len(part) / 2):
self.survived = False
def update_radius(self, part):
### Set radius as equal to length of furthest nn
# Get index of the furthest nn coordinate in norm_arr
mx = part[:, 3].max()
index = np.where(self.norm_arr == mx)[0][0]
# Get the pixel equivalent
x1, y1 = self.pixels[index, :2]
## euclidian distance
x2, y2 = self.pixel_mid
self.pixel_radius = m.sqrt((x2 - x1)**2 + (y2 - y1)**2)
""" Run once """
def init_k(self):
### k should be odd so you can make a clear prediction
k = int(m.sqrt(self.max_k))
if k % 2 == 0:
k += 1
return k
