Introduction to scikit-learn

[2014-09-12 Fri 19:55] speaker: Shawn Tan, RA for Speech Recognition Lab event: Friday Hacks, NUS Hackers ** Titanic dataset from Kaggel ** Load it with Pandas ** Filling missing values imputation ** Converting categorical values to numerical one hot encoding, split categorical field into n binary fields use DictVectorizer ** Fit and then transform for data transformations ** Fit and then predict for model prediction score to do a series of predictions ** Passenger class is a number but may be better interpreted as a categorical value ** Pipelines API pipeline = Pipeline([

• (‘dictionary’, DictVectorizer()),
• (‘predictor’, log_reg) ]) ** cross validation with KFold KFold returns pairs of arrays indicating the partitions ** visualizing logistic regression as finding a separating plane, similar to SVM ** related projects rapidminder - GUI eureqa desktop