|
import numpy as np |
|
import matplotlib.pyplot as plt |
|
from sklearn.linear_model import MultiTaskLasso, Lasso |
|
import gradio as gr |
|
import time |
|
|
|
rng = np.random.RandomState(42) |
|
|
|
|
|
def make_plot(n_samples, n_features, n_tasks, n_relevant_features, alpha, progress=gr.Progress()): |
|
|
|
progress(0, desc="Starting...") |
|
time.sleep(1) |
|
for i in progress.tqdm(range(100)): |
|
time.sleep(0.1) |
|
|
|
coef = np.zeros((n_tasks, n_features)) |
|
times = np.linspace(0, 2 * np.pi, n_tasks) |
|
for k in range(n_relevant_features): |
|
coef[:, k] = np.sin((1.0 + rng.randn(1)) * times + 3 * rng.randn(1)) |
|
|
|
X = rng.randn(n_samples, n_features) |
|
Y = np.dot(X, coef.T) + rng.randn(n_samples, n_tasks) |
|
|
|
coef_lasso_ = np.array([Lasso(alpha=0.5).fit(X, y).coef_ for y in Y.T]) |
|
coef_multi_task_lasso_ = MultiTaskLasso(alpha=alpha).fit(X, Y).coef_ |
|
|
|
fig = plt.figure(figsize=(8, 5)) |
|
|
|
feature_to_plot = 0 |
|
fig = plt.figure() |
|
lw = 2 |
|
plt.plot(coef[:, feature_to_plot], color="seagreen", linewidth=lw, label="Ground truth") |
|
plt.plot( |
|
coef_lasso_[:, feature_to_plot], color="cornflowerblue", linewidth=lw, label="Lasso" |
|
) |
|
plt.plot( |
|
coef_multi_task_lasso_[:, feature_to_plot], |
|
color="gold", |
|
linewidth=lw, |
|
label="MultiTaskLasso", |
|
) |
|
|
|
plt.legend(loc='upper center', bbox_to_anchor=(0.5, -0.05), |
|
ncol=3, fancybox=True, shadow=True) |
|
plt.axis("tight") |
|
plt.ylim([-1.1, 1.1]) |
|
fig.suptitle("Lasso, MultiTaskLasso and Ground truth time series") |
|
return fig |
|
|
|
|
|
model_card = f""" |
|
## Description |
|
Multi-task Lasso allows us to jointly fit multiple regression problems by enforcing the selected features to be the same across tasks. This example simulates sequential measurement. |
|
Each task is a time instant, and the relevant features, while being the same, vary in amplitude over time. Multi-task lasso imposes that features that are selected at one time point |
|
are selected for all time points. This makes feature selection more stable than by regular Lasso. |
|
## Model |
|
currentmodule: sklearn.linear_model |
|
class:`Lasso` and class: `MultiTaskLasso` are used in this example. |
|
Plots represent Lasso, MultiTaskLasso and Ground truth time series |
|
""" |
|
|
|
with gr.Blocks(theme=gr.themes.Glass(primary_hue=gr.themes.colors.gray, |
|
secondary_hue=gr.themes.colors.sky, |
|
text_size=gr.themes.sizes.text_lg), |
|
css=".gradio-container {background-color: #9ea9a9 }") as demo: |
|
|
|
gr.Markdown(''' |
|
<div> |
|
<h1 style='text-align: center'> Joint feature selection with multi-task Lasso </h1> |
|
</div> |
|
''') |
|
gr.Markdown(model_card) |
|
gr.Markdown("Original example Author: Alexandre Gramfort <alexandre.gramfort@inria.fr>") |
|
gr.Markdown( |
|
"Iterative conversion by: <a href=\"https://www.deamarialeon.com\">Dea María Léon</a>" |
|
) |
|
gr.Markdown("### Please select values and click submit:") |
|
|
|
with gr.Row().style(equal_height=True): |
|
n_samples = gr.Slider(50,500,value=100,step=50,label='Number of samples') |
|
n_features = gr.Slider(5,50,value=30,step=5,label='Features') |
|
n_tasks = gr.Slider(5,50,value=40,step=5,label='Tasks') |
|
n_relevant_features = gr.Slider(1,10,value=5,step=1,label='Relevant features') |
|
alpha = gr.Slider(0,10,value=1.0,step=0.5,label='Alpha Range') |
|
|
|
btn = gr.Button(value = 'Submit') |
|
|
|
btn.click(make_plot,inputs=[n_samples,n_features, n_tasks, n_relevant_features, alpha],outputs=[gr.Plot()]) |
|
|
|
demo.queue().launch() |
