Scikit-learn: The knowledge points used in the actual project (summary)

0. All projects are common:

http://blog.csdn.net/mmc2015/article/details/46851245 (DataSet format and Predictor)

http://blog.csdn.net/mmc2015/article/details/46852755 (load your own raw data)

(Entire corpus loaded for text categorization issues )

http://blog.csdn.net/mmc2015/article/details/46906409 (5. Load the built-in common data)

(Many common data sets are loaded , 5.) Dataset loading utilities)

http://blog.csdn.net/mmc2015/article/details/46705983 (Choosing the Right estimator (your question is suitable for what estimator to model it))

( a picture tells you, your question choose what estimator good , no longer have to try again)

http://blog.csdn.net/mmc2015/article/details/46857949 (training classifier, predictive new data, evaluation classifier)

http://blog.csdn.net/mmc2015/article/details/46858009 (using "Pipeline" unified Vectorizer = Transformer = classifier, Grid Search Assistant)

First, the text classification to use:

http://blog.csdn.net/mmc2015/article/details/46857887 (extracting features from a text file (TF, IDF))

(countvectorizer,tfidftransformer)

http://blog.csdn.net/mmc2015/article/details/46866537 (what Countvectorizer extracted TF did)

( in-depth interpretation of what Countvectorizer has done, directing us to do personalized preprocessing )

http://blog.csdn.net/mmc2015/article/details/46867773 (2.5.2. Implementing LSA via TRUNCATEDSVD (implicit semantic analysis))

(LSA,LDA analysis )

(Non-Scikit-learn) http://blog.csdn.net/mmc2015/article/details/46940373 (textanalytics) (1): Two types of Word relations- -paradigmatic vs. Syntagmatic)

(Non-Scikit-learn) http://blog.csdn.net/mmc2015/article/details/46941367 (textanalytics) (1): Two types of Word relations- -paradigmatic vs. Syntagmatic (continued))

(Lexical granularity Relationship: paradigmatic ( aggregation relationship: The same nature can be substituted for each other, using TFIDF-based similarity mining ) vs. Syntagmatic ( Combinatorial relationship: Co-occurrence, mutual information mining ))

(Non-Scikit-learn) http://blog.csdn.net/mmc2015/article/details/46771791 (Feature selection method (TF-IDF, Chi, and IG))

(Introduction of TF-IDF in feature selection, CHI Square and information gain in feature selection )

Second, data preprocessing used (4. Dataset transformations):

http://blog.csdn.net/mmc2015/article/details/46991465 (4.1. Pipeline and Featureunion:combining estimators (features combined with predictors; characteristics and characteristics))

(Combination of feature and Predictor, feature and feature )

http://blog.csdn.net/mmc2015/article/details/46992105 (4.2. Feature extraction (Feature extraction, not feature selection))

(loading features form dicts, feature hashing,text feature extraction, image feature extraction)

http://blog.csdn.net/mmc2015/article/details/46997379 (4.2.3. Text feature Extraction)

(text feature extraction)

http://blog.csdn.net/mmc2015/article/details/47016313 (4.3. preprocessing data (Standardi/normali/binari. zation, encoding, missing value))

(standardization, or mean removal and variance scaling (normalization: de-mean, except variance),normalization(normalized),Feature binarization(binary),Encoding categorical features(encoding category feature), imputation ofmissing values (attribution missing value))

http://blog.csdn.net/mmc2015/article/details/47066239 (4.4. Unsupervised dimensionality reduction (descending dimension))

(PCA, Random projections, Feature agglomeration (feature aggregation))

http://blog.csdn.net/mmc2015/article/details/47069869 (4.8. Transforming the prediction target (y))

(Label binarization,lable encoding (transform non-numerical labels to numerical labels))

Third, other important points of knowledge:

http://blog.csdn.net/mmc2015/article/details/46867597 (2.5. Matrix factor decomposition problem)

Copyright NOTICE: This article for Bo Master original article, without Bo Master permission not reproduced.

Scikit-learn: The knowledge points used in the actual project (summary)