导语

  

内容提要

  

    机器学习所涉及的复杂统计学知识困扰了很多开发者。知晓统计学知识可以帮助你为给定的问题构建强壮的机器学习优化模型。
    普拉塔普·丹格迪著的《机器学习统计学(影印版)(英文版)》将教你机器学习所需的实现复杂统计计算的相关内容，可以从中获得监督学习、无监督学习、强化学习等背后的统计学知识。你将看到讨论机器学习相关统计学内容的真实案例并熟悉它们。还能学到用于实现建模、调参、回归、分类、密度采集、向量处理、矩阵等的相关程序。
    学完本书，你会掌握机器学习所需的统计学知识。并且能够将所学新技能应用于任何行业问题。

Preface
Chapter 1: Journey from Statistics to Machine Learning
  Statistical terminology for model building and validation
  Machine learning
  Major differences between statistical modeling and machine learning
  Steps in machine learning model development and deployment
  Statistical fundamentals and terminology for model building andvalidation
  Bias versus variance trade-off
  Train and test data
  Machine learning terminology for model building and validation
  Linear regression versus gradient descent
  Machine learning losses
  When to stop tuning machine learning models
  Train, validation, and test data
  Cross-validation
  Grid search
  Machine learning model overview
  Summary
Chapter 2: Parallelism of Statistics and Machine Learning
  Comparison between regression and machine learning models
  Compensating factors in machine learning models
  Assumptions of linear regression
  Steps applied in linear regression modeling
  Example of simple linear regression from first principles
  Example of simple linear regression using the wine quality data
  Example of multilinear regression - step-by-step methodology of model
  building
  Backward and forward selection
  Machine learning models - ridge and lasso regression
  Example of ridge regression machine learning
  Example of lasso regression machine learning model
  Regularization parameters in linear regression and ridge/lasso regression
  Summary
Chapter 3: Logistic Regression Versus Random Forest
  Maximum likelihood estimation
  Logistic regression - introduction and advantages
  Terminology involved in logistic regression
  Applying steps in logistic regression modeling
  Example of logistic regression using German credit data
  Random forest
  Example of random forest using German credit data
  Grid search on random forest
  Variable importance plot
  Comparison of logistic regression with random forest
  Summary
Chapter 4: Tree-Based Machine Learning Models
  Introducing decision tree classifiers
  Terminology used in decision trees
  Decision tree working methodology from first principles
  Comparison between logistic regression and decision trees
  Comparison of error components across various styles of models
  Remedial actions to push the model towards the ideal region
  HR attrition data example
  Decision tree classifier
  Tuning class weights in decision tree classifier
  Bagging classifier
  Random forest classifier
  Random forest classifier - grid search
  AdaBoost classifier
  Gradient boosting classifier
  Comparison between AdaBoosting versus gradient boosting
  Extreme gradient boosting - XGBoost classifier
  Ensemble of ensembles - model stacking
  Ensemble of ensembles with different types of classifiers
  Ensemble of ensembles with bootstrap samples using a single type of
  classifier
  Summary
Chapter 5: K-Nearest Neighbors and Naive Bayes
  K-nearest neighbors
  KNN voter example
  Curse of dimensionality
  Curse of dimensionality with 1D, 2D, and 3D example
  KNN classifier with breast cancer Wisconsin data example
  Tuning of k-value in KNN classifier
  Naive Bayes
  Probability fundamentals
  Joint probability
  Understanding Bayes theorem with conditional probability
  Naive Bayes classification
  Laplace estimator
  Naive Bayes SMS spam classification example
  Summary
Chapter 6: Support Vector Machines and Neural Networks
  Support vector machines working principles
  Maximum margin classifier
  Support vector classifier
  Support vector machines
  Kernel functions
  SVM multilabel classifier with letter recognition data example
  Maximum margin classifier - linear kernel
  Polynomial kernel
  RBF kernel
  Artificial neural networks -ANN
  Activation functions
  Forward propagation and backpropagation
  Optimization of neural networks
  Stochastic gradient descent - SGD
  Momentum
  Nesterov accelerated gradient - NAG
  Adagrad
  Adadelta
  RMSprop
  Adaptive moment estimation - Adam
  Limited-memory broyden-fletcher-goldfarb-shanno - L-BFGS
  optimization algorithm
  Dropout in neural networks
  ANN classifier applied on handwritten digits using scikit-learn
  Introduction to deep learning
  Solving methodology
  Deep learning software
  Deep neural network classifier applied on handwritten digits using Keras
  Summary
Chapter 7: Recommendation Engines
  Content-based filtering
  Cosine similarity
  Collaborative filtering
  Advantages of collaborative filtering over content-based filtering
  Matrix factorization using the alternating least squares algorithm for
  collaborative filtering
  Evaluation of recommendation engine model
  Hyperparameter selection in recommendation engines using grid search
  Recommendation engine application on movie lens data
  User-user similarity matrix
  Movie-movie similarity matrix
  Collaborative filtering using ALS
  Grid search on collaborative filtering
  Summary
Chapter 8: Unsupervised Learning
  K-means clustering
  K-means working methodology from first principles
  Optimal number of clusters and cluster evaluation
  The elbow method
  K-means clustering with the iris data example
  Principal component analysis - PCA
  PCA working methodology from first principles
  PCA applied on handwritten digits using scikit-learn
  Singular value decomposition - SVD
  SVD applied on handwritten digits using scikit-learn
  Deep auto encoders
  Model building technique using encoder-decoder architecture
  Deep auto encoders applied on handwritten digits using Keras
  Summary
Chapter 9: Reinforcement Learning
  Introduction to reinforcement learning
  Comparing supervised, unsupervised, and reinforcement learning in detail
  Characteristics of reinforcement learning
  Reinforcement learning basics
  Category 1 - value based
  Category 2 - policy based
  Category 3 - actor-critic
  Category 4 - model-free
  Category 5 - model-based
  Fundamental categories in sequential decision making
  Markov decision processes and Bellman equations
  Dynamic programming
  Algorithms to compute optimal policy using dynamic programming
  Grid world example using value and policy iteration algorithms with basic Python
  Monte Carlo methods
  Comparison between dynamic programming and Monte Carlo methods
  Key advantages of MC over DP methods
  Monte Carlo prediction
  The suitability of Monte Carlo prediction on grid-world problems
  Modeling Blackjack example of Monte Carlo methods using Python
  Temporal difference learning
  Comparison between Monte Carlo methods and temporal difference
  learning
  TD prediction
  Driving office example for TD learning
  SARSA on-policy TD control
  Q-learning - off-policy TD control
  Cliff walking example of on-policy and off-policy of TD control
  Applications of reinforcement learning with integration of machine
  learning and deep learning
  Automotive vehicle control - self-driving cars
  Google DeepMind's AlphaGo
  Robo soccer
  Further reading
  Summary
Index