Spaces:
Runtime error
Runtime error
File size: 9,627 Bytes
71ff1f9 af0b767 71ff1f9 967ed47 71ff1f9 967ed47 71ff1f9 af0b767 967ed47 71ff1f9 af0b767 71ff1f9 967ed47 71ff1f9 967ed47 71ff1f9 967ed47 71ff1f9 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 |
from sklearn import datasets, ensemble
from sklearn.inspection import permutation_importance
from sklearn.metrics import mean_squared_error
from sklearn.model_selection import train_test_split
import plotly.graph_objs as go
import numpy as np
import plotly.express as px
import pandas as pd
import gradio as gr
diabetes = datasets.load_diabetes(as_frame=True)
X, y = diabetes.data, diabetes.target
def display_table(row_number):
X, y = diabetes.data, diabetes.target
XX = pd.concat([X, y], axis=1)
temp_df = XX[row_number : row_number + 5]
Statement = f"Displaying rows from row {row_number} to {row_number+5}"
return Statement, temp_df
def train_model(
test_split,
learning_rate,
n_estimators,
max_depth,
min_samples_split,
loss,
duration,
):
X, y = diabetes.data, diabetes.target
X_train, X_test, y_train, y_test = train_test_split(
X, y, test_size=test_split, random_state=42
)
params = {
"n_estimators": n_estimators,
"max_depth": max_depth,
"min_samples_split": min_samples_split,
"learning_rate": learning_rate,
"loss": loss,
}
global reg
reg = ensemble.GradientBoostingRegressor(**params)
reg.fit(X_train, y_train)
mse = mean_squared_error(y_test, reg.predict(X_test))
x = np.arange(params["n_estimators"]) + 1
train_score = reg.train_score_
test_score = np.zeros((params["n_estimators"],), dtype=np.float64)
for i, y_pred in enumerate(reg.staged_predict(X_test)):
test_score[i] = mean_squared_error(y_test, y_pred)
test_score = test_score
fig = go.Figure()
fig.add_trace(
go.Scatter(
x=x,
y=train_score,
mode="lines",
name="Training Set Deviance",
line=dict(color="blue"),
)
)
fig.add_trace(
go.Scatter(
x=x,
y=test_score,
mode="lines",
name="Test Set Deviance",
line=dict(color="red"),
)
)
frames = [
go.Frame(
data=[
go.Scatter(
x=x[: k + 1],
y=train_score[: k + 1],
mode="lines",
line=dict(color="blue"),
),
go.Scatter(
x=x[: k + 1],
y=test_score[: k + 1],
mode="lines",
line=dict(color="red"),
),
],
name=f"frame{k}",
)
for k in range(1, len(x))
]
fig.frames = frames
fig.update_layout(
title="Deviance",
xaxis_title="Boosting Iterations",
yaxis_title="Deviance",
legend=dict(x=0, y=1),
updatemenus=[
dict(
type="buttons",
showactive=False,
direction="right",
pad={"r": 10},
buttons=[
dict(
label="Play",
method="animate",
args=[
None,
dict(
frame=dict(duration=duration, redraw=True),
fromcurrent=True,
transition=dict(duration=0),
),
],
),
dict(
label="Pause",
method="animate",
args=[
[None],
dict(
frame=dict(duration=0, redraw=False),
mode="immediate",
transition=dict(duration=0),
),
],
),
],
x=0.5,
y=-0.2,
)
],
)
return fig, mse
def Plot_featue_importance(test_split):
try:
feature_importance = reg.feature_importances_
except:
# return blank figures
fig = go.Figure()
fig.update_layout(title="Train a model to see the plots")
return fig, fig
sorted_idx = np.argsort(feature_importance)
fig = px.bar(
pd.DataFrame(
{
"Importance": feature_importance[sorted_idx],
"Feature": np.array(diabetes.feature_names)[sorted_idx],
}
),
x="Importance",
y="Feature",
orientation="h",
title="Feature Importance (MDI)",
)
X, y = diabetes.data, diabetes.target
X_train, X_test, y_train, y_test = train_test_split(
X, y, test_size=test_split, random_state=42
)
result = permutation_importance(
reg, X_test, y_test, n_repeats=10, random_state=42, n_jobs=2
)
fig1 = px.box(
pd.DataFrame(result.importances.T, columns=diabetes.feature_names),
title="Permutation Importance (test set)",
)
return fig, fig1
with gr.Blocks() as demo:
gr.Markdown("# Gradient Boosting regression")
gr.Markdown(
"This demo is based on [gradient boosting regression example of scikit-learn](https://scikit-learn.org/stable/auto_examples/ensemble/plot_gradient_boosting_regression.html) Example.This example demonstrates gradient goosting to produce a predictive model from an ensemble of weak predictive models. Gradient boosting can be used for regression and classification problems. Here, we will train a model to tackle a diabetes regression task."
)
with gr.Tab("Train the model"):
gr.Markdown("### Below is the diabetes dataset used in this demo 👇 ")
gr.Markdown("### You can change the interval of rows to display.")
gr.Markdown(
"The diabetes dataset consists of ten baseline variables, age, sex, body mass index (BMI), average blood pressure (BP), and six blood serum measurements for 442 diabetes patients. The target variable is a quantitative measure of disease progression one year after baseline."
)
total_rows = X.shape[0]
rows_number = gr.Slider(
0, total_rows, label="Displaying Rows", value=5, step=5
)
rows_number.change(
fn=display_table,
inputs=[rows_number],
outputs=[gr.Text(label="Row"), gr.DataFrame()],
)
gr.Markdown(
"# Play with the parameters to see how the model performance changes"
)
gr.Markdown(
"""
### `Number of Estimators` : the number of boosting stages that will be performed. Later, we will plot deviance against boosting iterations.
### `Max Depth` : limits the number of nodes in the tree. The best value depends on the interaction of the input variables.
### `Min Samples Split` : the minimum number of samples required to split an internal node.
### `learning_rate` : how much the contribution of each tree will shrink.
### `loss` : loss function to optimize.
### `Test Split` : the percentage of the dataset to include in the test split.
### `Animation Speed for Deviance Plot` : the duration of the animation of Deviation Plot.
"""
)
with gr.Row():
test_split = gr.Slider(0.1, 0.9, label="Test Split", value=0.2, step=0.1)
learning_rate = gr.Slider(
0.01, 0.5, label="Learning Rate", value=0.1, step=0.01
)
n_estimators = gr.Slider(
10, 1000, label="Number of Estimators", value=100, step=10
)
max_depth = gr.Slider(1, 10, label="Max Depth", value=3, step=1)
min_samples_split = gr.Slider(
2, 10, label="Min Samples Split", value=2, step=1
)
loss = gr.Dropdown(
["squared_error", "absolute_error", "huber", "quantile"],
label="Loss",
value="squared_error",
)
duration = gr.Slider(
0, 100, label="Animation Speed for Deviance Plot", value=25, step=10
)
model_btn = gr.Button("Train Model")
gr.Markdown(
"### Finally, we will visualize the results. To do that, we will first compute the test set deviance and then plot it against boosting iterations."
)
model_btn.click(
fn=train_model,
inputs=[
test_split,
learning_rate,
n_estimators,
max_depth,
min_samples_split,
loss,
duration,
],
outputs=[gr.Plot(), gr.Text(label="MSE")],
)
with gr.Tab("Feature Importance"):
gr.Markdown("## Feature Importance (MDI) and Permutation Importance (test set)")
gr.Markdown(
"For this example, the impurity-based and permutation methods identify the same 2 strongly predictive features but not in the same order. The third most predictive feature, “bp”, is also the same for the 2 methods. The remaining features are less predictive and the error bars of the permutation plot show that they overlap with 0."
)
feat_imp_btn = gr.Button("Plot Feature Importance")
with gr.Row():
feat_imp_btn.click(
fn=Plot_featue_importance,
inputs=[test_split],
outputs=[gr.Plot(), gr.Plot()],
)
demo.launch()
|