[ 346 ]
using, for nonlinear dimensionality reduction 89,
90, 91
L
label propagation
about 312
with semi-supervised learning 312, 313, 314
LASSO cross-validation 124
least absolute shrinkage and selection operator
(LASSO)
about 123
for feature selection 125
least angle regression (LARS) 123
leave-one-out cross-validation (LOOCV) 119
line
fitting, through data 105, 106, 107
linear algorithm
versus nonlinear algorithm 40
linear discriminant analysis (LDA)
using, for classification 299, 300, 302, 304
linear methods
using, for classification 131
linear regression model
evaluating 110, 111, 113
logistic regression errors
examining, with confusion matrix 139
logistic regression
about 28, 131
classification threshold, varying in 141, 143, 144
M
machine learning algorithms
interpretability 41
machine learning linear regression
versus traditional linear regression 130
machine learning, with logistic regression
about 137
features, defining 138
logistic regression, scoring 139
logistic regression, training 138
target arrays, defining 138
testing set, providing 138
training set, providing 138
machine learning
line, fitting through data 108, 109
overview 36
manifolds
dimensionality reduction, performing with 98,
100
matplotlib
plotting with 21, 22, 24
maximum a posteriori (MAP) 310
mean absolute deviation (MAD) 112
mean squared error (MSE) 112
MiniBatch k-means
used, for handling data 170, 172
missing values
imputing, through strategies 59, 60, 61
models
persisting, with joblib 230
persisting, with pickle 230
regularizing, sparsity used 123
selecting, with cross-validation 196, 197, 198
multi-dimensional scaling (MDS) 101
multiclass SVC classifier 244
N
Naive Bayes
documents, classifying 307, 308, 310
NaN values 14
natural language processing (NLP) 102
negative predictive value (NPV) 144
neural network
first base model 326
philosophical thoughts 324
stacking with 324, 325
using, in scikit-learn 321, 322, 323
non-negative matrix factorization (NMF) 102
nonlinear algorithm
versus linear algorithm 40
nonlinear dimensionality reduction
kernel PCA, using for 89, 90, 91
nonlinear LDA 304
NumPy arrays
dimension 9
initializing 12
shape 9
NumPy
basics 8
broadcasting 10, 11