def todo_vs_gamma():
with open('csv/todo2.csv', 'w') as f:
f.write('hitrate,precision,componentes')
f.write('\n')
for g in [100, 120]:
kfold.generate(g)
r = execute(1)
m = confusion_matrix(r)
hr, pr = stats_avg(m)
f.write('{},{},{}'.format(hr, pr, g))
f.write('\n')
def gamma_vs_k():
with open('csv/hitrates.csv', 'w') as f:
f.write('hitrate,componentes,k')
f.write('\n')
for g in [1, 5, 10, 15]:
for k in np.arange(1, 20, 2):
kfold.generate(g)
r = execute(k)
m = confusion_matrix(r)
hr, pr = stats_avg(m)
f.write('{},{},{}'.format(hr, g, k))
f.write('\n')
def gamma_vs_time():
with open('csv/times.csv', 'w') as f:
f.write('time,componentes')
f.write('\n')
for g in np.arange(1, 40, 4):
kfold.generate(g)
r = execute(1)
start = time.time()
list(r)
end = time.time()
f.write('{},{}'.format(end - start, g))
f.write('\n')
def stats_precision():
with open('csv/precision.csv', 'w') as f:
f.write('precision,componentes,k')
f.write('\n')
diag = np.arange(personas)
for g in [15, 25, 35]:
for k in [1, 5, 10]:
kfold.generate(g)
r = execute(k)
m = confusion_matrix(r)
precision = m[diag, diag] / np.sum(m, axis=1)
for p in precision:
f.write('{},{},{}'.format(p, g, k))
f.write('\n')
def stats_hitrate():
diag = np.arange(personas)
with open('csv/hitrate_ambos.csv', 'w') as f:
f.write('imagenes,hitrate,componentes,k')
f.write('\n')
diag = np.arange(personas)
for g in [15, 25, 35]:
for k in [1, 5, 10]:
kfold.generate(g)
r = execute(k)
m = confusion_matrix(r)
hitrate = m[diag, diag] / imagenes
for h in hitrate:
f.write('{},{},{},{}'.format(imagenes - fold_size, h, g,
k))
f.write('\n')