Shivangi Patel CrazyDaffodils

CrazyDaffodils / data_loader.py

Created September 1, 2020 20:36 — forked from kevinzakka/data_loader.py

Train, Validation and Test Split for torchvision Datasets

	"""
	Create train, valid, test iterators for CIFAR-10 [1].
	Easily extended to MNIST, CIFAR-100 and Imagenet.

	[1]: https://discuss.pytorch.org/t/feedback-on-pytorch-for-kaggle-competitions/2252/4
	"""

	import torch
	import numpy as np

CrazyDaffodils / UMAP

Created January 14, 2020 00:23

	from umap import UMAP
	import seaborn as sns
	model = UMAP(n_neighbors = 40, min_dist = 0.4, n_components = 2)
	umap = model.fit_transform(X_std)
	umap_df = pd.DataFrame(data=umap, columns=['UMAP1','UMAP2']).join(labels)
	palette = sns.color_palette("muted", n_colors=5)
	sns.set_style("white")
	sns.scatterplot(x='UMAP1',y='UMAP2',hue='Class',data=umap_df, palette=palette, linewidth=0.2, s=30,
	alpha=1).set_title('UMAP')

CrazyDaffodils / t-SNE

Last active January 13, 2020 23:43

	from sklearn.manifold import TSNE
	import seaborn as sns
	#Visualize data using t-SNE.
	model = TSNE(learning_rate = 10, n_components = 2, random_state=123, perplexity = 30)
	tsne = model.fit_transform(X_std)
	tsne_df = pd.DataFrame(data=tsne, columns=['t-SNE1','t-SNE2']).join(labels)
	palette = sns.color_palette("muted", n_colors=5)
	sns.set_style("white")
	sns.scatterplot(x='t-SNE1',y='t-SNE2',hue='Class',data=tsne_df, palette=palette, linewidth=0.2, s=30, alpha=1).set_title('t-SNE')

CrazyDaffodils / pca_explained variance

Last active January 13, 2020 23:39

	from sklearn.decomposition import PCA
	import matplotlib.pyplot as plt
	pca_std = PCA().fit(X_std)
	percent_variance=pca_std.explained_variance_ratio_*100
	plt.figure()
	plt.plot(np.cumsum(pca_std.explained_variance_ratio_))
	plt.xlabel('Number of Components')
	plt.ylabel('Variance (%)') #for each component
	plt.show()

CrazyDaffodils / pca

Last active January 13, 2020 23:33

	from sklearn.decomposition import PCA
	import seaborn as sns
	#Visualize data using Principal Component Analysis.
	print("Principal Component Analysis (PCA)")
	pca = PCA(n_components = 2).fit_transform(X_std)
	pca_df = pd.DataFrame(data=pca, columns=['PC1','PC2']).join(labels)
	palette = sns.color_palette("muted", n_colors=5)
	sns.set_style("white")
	sns.scatterplot(x='PC1',y='PC2',hue='Class',data=pca_df, palette=palette, linewidth=0.2, s=30, alpha=1).set_title('PCA')