/Unsupervised-Learning

Clustering by scikit-learn

Primary LanguageJupyter Notebook

Unsupervised Learning - Clustering

用scikit-learn執行unsupervised learning的基礎，包含以下程式碼。

1 Unsupervised Learning.ipynb

此範例說明如何將高維度的特徵降維以顯示在圖上，方法包含Principal Components Analysis(PCA)、Multidimensional scaling (MDS)、t-SNE。
並說明如何用scikit-learn做clustering，方法包含K-means、Agglomerative clustering、DBSCAN clustering。

2 text clustering.ipynb

此範例說明如何對未標簽的文章做分類，首先用TF-IDF(Term Frequency - Inverse Document Frequency)提取特徵，再用K-Means做分群。

3 BBC news clustering.ipynb

此範例和上一例類似，做資料量較大的新聞分群。

4 mnist auto encoder.ipynb

此範例用autoencoder(一種self-supervised learning)，做MNIST的分群。
image4

TF-IDF

代表文章特徵的一種方法。
image1
image2
image3

References

https://scikit-learn.org/stable/modules/generated/sklearn.cluster.KMeans.html
https://www.coursera.org/learn/python-machine-learning?specialization=data-science-python
https://www.coursera.org/learn/ml-foundations?specialization=data-science-python
https://www.kaggle.com/sameersmahajan/people-wikipedia-data
https://github.com/aspamers/autoencoder
https://blog.keras.io/building-autoencoders-in-keras.html
https://machinelearningmastery.com/lstm-autoencoders/