在职个人战拖打卡DataCamp

06/21 是时候做个了结了！打卡完成Cluster Analysis in Python～

Performance-wise: K-means > hierarchy clustering

K-means
1. Generate cluster centers: kmeans(), returns centroid and distortion (sum of squares of distances from the centroid)
2. Generate cluster labels: vq(), returns a list of cluster labels and a list of distortions

Image clustering
1. extract RGB values in three lists
2. cluster based on scaled RGB values (value/std(color)*255)
3. display the colors of cluster centers

Document clustering
1. Create a sparse matrix with Tfidf Vectorizer: TfidfVectorizer(max_df = , min_df = , max_featutres = , tokenizer = ); .fit_transform()
2. Cluster the tfidf matrix (use .todense() to convert sparse matrix to matrics) and get the list of terms: tfidf_model.get_feature_names()