classification.py 文件源码

def reduceDimensionality(X, y, method="selectkbest", targetDim=10):
    """ Reduces the dimensionality of [X] to [targetDim] """
    try:
        # Check for the required methodology first
        if method.lower() == "selectkbest":
            prettyPrint("Selecting %s best features from dataset" % targetDim, "debug")
            kBestSelector = SelectKBest(k=targetDim)
            X_new = kBestSelector.fit_transform(X, y).tolist()
        elif method.lower() == "pca":
            prettyPrint("Extracting %s features from dataset using PCA" % targetDim, "debug")
            pcaExtractor = PCA(n_components=targetDim)
            # Make sure vectors in X are positive
            X_new = pcaExtractor.fit_transform(X, y).tolist()
        else:
            prettyPrint("Unknown dimensionality reduction method \"%s\"" % method, "warning")
            return X

    except Exception as e:
        prettyPrint("Error encountered in \"reduceDimensionality\": %s" % e, "error")
        return X

    # Return the reduced dataset
    return X_new