1 一个热向量应该用数值属性缩放吗 在具有分类属性和数字属性的组合的情况下,我通常将分类属性转换为一个热向量。我的问题是我应该保留这些向量不变,并通过标准化/归一化来缩放数字属性,还是应该将一个热向量与数字属性一起缩放? 20 feature-engineering feature-scaling data-science-model
3 是否有适用于python的好的即用型语言模型? 我正在为一个应用程序制作原型,我需要一个语言模型来计算一些生成的句子的困惑度。 我可以随时使用经过训练的python语言模型吗?简单的东西 model = LanguageModel('en') p1 = model.perplexity('This is a well constructed sentence') p2 = model.perplexity('Bunny lamp robert junior pancake') assert p1 < p2 我看过一些框架,但找不到我想要的。我知道我可以使用类似: from nltk.model.ngram import NgramModel lm = NgramModel(3, brown.words(categories='news')) 这在Brown Corpus上使用了很好的图林概率分布,但是我正在一些大型数据集(例如1b单词数据集)上寻找精心设计的模型。我可以真正相信一般领域的结果(不仅是新闻) 11 python nlp language-model r statistics linear-regression machine-learning classification random-forest xgboost python sampling data-mining orange predictive-modeling recommender-system statistics dimensionality-reduction pca machine-learning python deep-learning keras reinforcement-learning neural-network image-classification r dplyr deep-learning keras tensorflow lstm dropout machine-learning sampling categorical-data data-imputation machine-learning deep-learning machine-learning-model dropout deep-network pandas data-cleaning data-science-model aggregation python neural-network reinforcement-learning policy-gradients r dataframe dataset statistics prediction forecasting r k-means python scikit-learn labels python orange cloud-computing machine-learning neural-network deep-learning rnn recurrent-neural-net logistic-regression missing-data deep-learning autoencoder apache-hadoop time-series data preprocessing classification predictive-modeling time-series machine-learning python feature-selection autoencoder deep-learning keras tensorflow lstm word-embeddings predictive-modeling prediction machine-learning-model machine-learning classification binary theory machine-learning neural-network time-series lstm rnn neural-network deep-learning keras tensorflow convnet computer-vision