数据集获取

git clone https://github.com/Demon-2-Angel/Cereberal-Stroke-Analysis.git

curl -L -o repo.zip https://github.com/Demon-2-Angel/Cereberal-Stroke-Analysis/archive/refs/heads/main.zip
unzip repo.zip

源站 README 摘录（使用方式）

Cereberal-Stroke-Analysis

Followed Process

Read Data:

The script starts by importing necessary libraries (pandas, numpy, seaborn, matplotlib.pyplot) and reading a CSV file into a DataFrame (df).

Exploratory Data Analysis (EDA):

Basic exploration of the dataset using head(), describe(), and checking for missing values using isnull().sum().

Handling Categorical Variables:

One-hot encoding is performed on categorical variables using pd.get_dummies().

Handling Missing Values:

Missing values are imputed using the k-nearest neighbors algorithm (KNNImputer from sklearn.impute).

Feature Scaling and Train-Test Split:

Features are scaled using MinMaxScaler, and the dataset is split into training and testing sets.

Model Selection:

Several classification models are chosen (KNeighborsClassifier, GaussianNB, DecisionTreeClassifier, and RandomForestClassifier) for initial testing.

Model Evaluation Without Resampling:

Classification reports are generated for each model to evaluate their performance on the imbalanced dataset.

<p align=“center”>
<img src=“https://github.com/Demon-2-Angel/Cereberal-Stroke-Analysis/blob/main/Images/Before Sampling.png”>
</p>

OverSampling (SMOTE):

The script uses the Synthetic Minority Over-sampling Technique (SMOTE) to oversample the minority class.

Model Evaluation After OverSampling:

The same models are re-trained and evaluated on the oversampled dataset.

<p align=“center”>
<img src=“https://github.com/Demon-2-Angel/Cereberal-Stroke-Analysis/blob/main/Images/OverSampling.png”>
</p>

UnderSampling:

Random under-sampling is performed to balance the class distribution.

Model Evaluation After UnderSampling:

The models are re-trained and evaluated on the undersampled dataset.

<p align=“center”>
<img src=“https://github.com/Demon-2-Angel/Cereberal-Stroke-Analysis/blob/main/Images/UnderSampling.png”>
</p>

Combining OverSampling and UnderSampling (SMOTEENN):

The SMOTEENN technique, which combines SMOTE and Edited Nearest Neighbours (ENN), is applied.

Model Evaluation After Combining OverSampling and UnderSampling:

The models are re-trained and evaluated on the combined dataset.

<p align=“center”>
<img src=“https://github.com/Demon-2-Angel/Cereberal-Stroke-Analysis/blob/main/Images/After Over %26 Under Sampling.png”>
</p>

Conclusion:

数据加载示例（图像类）

from PIL import Image
import glob, os

files = (glob.glob(os.path.join(path, "**", "*.png"), recursive=True)
       + glob.glob(os.path.join(path, "**", "*.jpg"), recursive=True)
       + glob.glob(os.path.join(path, "**", "*.tif"), recursive=True))
print("图像文件数:", len(files))
img = Image.open(files[0]); print("尺寸/模式:", img.size, img.mode)

# torchvision Dataset 方式:
# from torchvision import datasets
# ds = datasets.ImageFolder(path)  # 要求 子目录=类别

目录组织与标注格式以源站说明和下载后实际文件为准。

完整仓库：github.com/Demon-2-Angel/Cereberal-Stroke-Analysis