def plot_training_curve(self, xticks_step=5):
evolution = pd.DataFrame({
'Generation': self.log.select("gen"),
'Max Accuracy': self.log.select("max"),
'Average Accuracy': self.log.select("avg"),
'Min Accuracy': self.log.select("min")
})
plt.title('Hyperparameter Optimisation')
plt.plot(evolution['Generation'],
evolution['Min Accuracy'],
'b',
color='C1',
label='Min')
plt.plot(evolution['Generation'],
evolution['Average Accuracy'],
'b',
color='C2',
label='Average')
plt.plot(evolution['Generation'],
evolution['Max Accuracy'],
'b',
color='C3',
label='Max')
plt.legend(loc='lower right')
plt.ylabel('Accuracy')
plt.xlabel('Generation')
plt.xticks(
[x for x in range(0, self.number_of_generations + 1, xticks_step)])
plt.show()
def fit(self):
self.X_embedded = TSNE(n_components=self.n_comp,
perplexity=self.perplexity,
verbose=self.verbose,
init=self.init).fit_transform(self.data)
self.df_embedded = pd.DataFrame(
{f'{n+1}d': self.X_embedded[:, n]
for n in range(self.n_comp)})
def get_centroids(self):
centroid_df = pd.DataFrame(self.X_projected)
centroid_df[self.category_label] = self.categories
df_centroids_projected = get_centroids_from_categories(
centroid_df, self.category_label)
X_centroids_projected = df_centroids_projected.values
centroids_categories = df_centroids_projected.index
return X_centroids_projected, centroids_categories
def fit(self):
self.pca.fit(self.X_scaled)
self.X_projected = self.pca.transform(self.X_scaled)
# Components table
components_cols = ["F{}".format(n + 1) for n in range(self.n_comp)]
self.components_table = pd.DataFrame(self.X_projected,
index=self.categories,
columns=components_cols)
self.default_factorial_plan_nb = self.kaiser_criterion(
pair_comp=True) / 2
def fit(self):
"""
"""
self.pca.fit(self.X_scaled)
self.X_projected = self.pca.transform(self.X_scaled)
self.evr = self.pca.explained_variance_ratio_
# Components table
components_cols = ["F{}".format(n + 1) for n in range(self.n_comp)]
self.components_table = pd.DataFrame(self.X_projected,
index=self.categories,
columns=components_cols)
self.default_factorial_plan_nb = int(self.kaiser_criterion() // 2)