d990f23b72c2409084a799bdf49109a996a02256,demo/PCA_demo.py,,,#,3

Before Change

from models import PCA as PCA
from utils import data_loader as data_loader

para = {
"path":"../../Data/SOSP_data/", // directory for input data
"log_seq_file_name":"rm_repeat_rawTFVector.txt", // filename for log sequence data file
"label_file_name":"rm_repeat_mlabel.txt", // filename for label data file
"fraction":0.95
}


if __name__ == "__main__":
	raw_data, label_data = data_loader.hdfs_data_loader(para)

After Change

    feature_extractor = preprocessing.FeatureExtractor()
    x_train = feature_extractor.fit_transform(x_train, term_weighting="tf-idf", 
                                            normalization="zero-mean")
    x_test = feature_extractor.transform(x_test)

    model = PCA()
    model.fit(x_train)

Instances