Delete app.py
Browse files
app.py
DELETED
@@ -1,112 +0,0 @@
|
|
1 |
-
# Authors: Jona Sassenhagen
|
2 |
-
# License: BSD 3 clause
|
3 |
-
|
4 |
-
import matplotlib.pyplot as plt
|
5 |
-
import numpy as np
|
6 |
-
from sklearn.decomposition import FactorAnalysis, PCA
|
7 |
-
from sklearn.preprocessing import StandardScaler
|
8 |
-
from sklearn.datasets import load_iris
|
9 |
-
import gradio as gr
|
10 |
-
from matplotlib import style
|
11 |
-
plt.switch_backend("agg")
|
12 |
-
style.use('ggplot')
|
13 |
-
|
14 |
-
font1 = {'family':'DejaVu Sans','color':'#2563EB','size': 14}
|
15 |
-
|
16 |
-
#load and transform the data
|
17 |
-
data = load_iris()
|
18 |
-
X = StandardScaler().fit_transform(data["data"])
|
19 |
-
feature_names = data["feature_names"]
|
20 |
-
|
21 |
-
methods = {
|
22 |
-
"PCA": PCA(),
|
23 |
-
"Unrotated FA": FactorAnalysis(),
|
24 |
-
"Varimax FA": FactorAnalysis(rotation="varimax")
|
25 |
-
}
|
26 |
-
|
27 |
-
def factor_analysis(method):
|
28 |
-
#figure1
|
29 |
-
fig1, ax = plt.subplots(figsize=(10, 6), facecolor='none', dpi = 200)
|
30 |
-
im = ax.imshow(np.corrcoef(X.T), cmap="Spectral", vmin=-1, vmax=1)
|
31 |
-
|
32 |
-
ax.set_xticks([0, 1, 2, 3])
|
33 |
-
ax.set_xticklabels(list(feature_names),
|
34 |
-
rotation=90, fontdict = font1)
|
35 |
-
ax.set_yticks([0, 1, 2, 3])
|
36 |
-
ax.set_yticklabels(list(feature_names), fontdict = font1)
|
37 |
-
plt.grid(False)
|
38 |
-
plt.colorbar(im).ax.tick_params()
|
39 |
-
ax.set_title("Iris feature correlation matrix",
|
40 |
-
fontdict=font1, size = 18,
|
41 |
-
color = "white", pad = 20,
|
42 |
-
bbox=dict(boxstyle="round,pad=0.3",
|
43 |
-
color = "#2563EB"))
|
44 |
-
plt.tight_layout()
|
45 |
-
plt.close('all')
|
46 |
-
|
47 |
-
n_comps = 2
|
48 |
-
|
49 |
-
#figure2
|
50 |
-
fig2, axs = plt.subplots(figsize=(8, 5), facecolor='none', dpi = 200)
|
51 |
-
plt.grid(False)
|
52 |
-
fa = methods[method]
|
53 |
-
fa.set_params(n_components=n_comps)
|
54 |
-
fa.fit(X)
|
55 |
-
|
56 |
-
components = fa.components_
|
57 |
-
|
58 |
-
vmax = np.abs(components).max()
|
59 |
-
axs.imshow(components, cmap="Spectral", vmax=vmax, vmin=-vmax)
|
60 |
-
axs.set_xticks(np.arange(len(feature_names)))
|
61 |
-
axs.set_xticklabels(feature_names, fontdict=font1)
|
62 |
-
axs.set_title(method,
|
63 |
-
fontdict=font1, size = 18,
|
64 |
-
color = "white", pad = 20,
|
65 |
-
bbox=dict(boxstyle="round,pad=0.3",
|
66 |
-
color = "#2563EB"))
|
67 |
-
axs.set_yticks([0, 1])
|
68 |
-
axs.set_yticklabels(["Comp. 1", "Comp. 2"], fontdict=font1)
|
69 |
-
|
70 |
-
plt.tight_layout()
|
71 |
-
plt.close('all')
|
72 |
-
|
73 |
-
return fig1, fig2, components
|
74 |
-
|
75 |
-
intro = """<h1 style="text-align: center;">🤗 <strong>Factor Analysis (with rotation) to visualize patterns</strong> 🤗</h1>
|
76 |
-
"""
|
77 |
-
desc = """<h3 style="text-align: left;"> Investigating the Iris dataset, we see that sepal length, petal length and petal width are highly correlated.
|
78 |
-
Sepal width is less redundant. Matrix decomposition techniques can uncover these latent patterns.
|
79 |
-
<br><br>Applying rotations to the resulting components does not inherently improve the predictive value of the derived latent space,
|
80 |
-
but can help visualise their structure; here, for example, the varimax rotation,
|
81 |
-
which is found by maximizing the squared variances of the weights,
|
82 |
-
finds a structure where the second component only loads positively on sepal width.
|
83 |
-
<br></h3>
|
84 |
-
"""
|
85 |
-
|
86 |
-
made ="""<div style="text-align: center;">
|
87 |
-
<p>Made with ❤</p>"""
|
88 |
-
|
89 |
-
link = """<div style="text-align: center;">
|
90 |
-
<a href="https://scikit-learn.org/stable/auto_examples/decomposition/plot_varimax_fa.html#sphx-glr-auto-examples-decomposition-plot-varimax-fa-py" target="_blank" rel="noopener noreferrer">
|
91 |
-
Demo is based on this script from scikit-learn documentation</a>"""
|
92 |
-
|
93 |
-
with gr.Blocks(theme=gr.themes.Soft(primary_hue="blue",
|
94 |
-
secondary_hue="sky",
|
95 |
-
neutral_hue="neutral",
|
96 |
-
font = gr.themes.GoogleFont("Roboto")),
|
97 |
-
title="Factor-Analysis-with-rotation") as demo:
|
98 |
-
gr.HTML(intro)
|
99 |
-
gr.HTML(desc)
|
100 |
-
method = gr.Radio(["PCA", "Unrotated FA", "Varimax FA"], label = "Choose method to show on the plot:", value = "PCA")
|
101 |
-
with gr.Box():
|
102 |
-
with gr.Column():
|
103 |
-
components = gr.Dataframe(headers= feature_names,label = "Loadings")
|
104 |
-
with gr.Row():
|
105 |
-
fig1 = gr.Plot(label="Plot covariance of Iris features")
|
106 |
-
fig2 = gr.Plot(label="Factor analysis")
|
107 |
-
method.change(fn=factor_analysis, inputs=method, outputs=[fig1, fig2, components])
|
108 |
-
demo.load(fn=factor_analysis, inputs=method, outputs=[fig1, fig2, components])
|
109 |
-
gr.HTML(made)
|
110 |
-
gr.HTML(link)
|
111 |
-
|
112 |
-
demo.launch()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|