Supervised Machine Learning, Regression and Classification

Supervised Machine Learning, Regression and Classification

1. Subject: Machine Learning Module I I : Supervised Learning 1

2. Unit- I Supervised Learning (Regression/Classification): Basic methods: Distance- based methods, Nearest- Neighbors, Decision Trees, Naive Bayes, Linear models: Linear Regression, Logistic Regression, Support Vector Machines, Nonlinearity and Kernel Methods, Beyond Binary Classification: Multi- class Unit- II Unsupervised Learning: Clustering: K- means, Dimensionality Reduction: PCA and kernel PCA, Generative Models (Gaussian Mixture Models and Hidden Markov Models) Unit-III Evaluating Machine Learning algorithms, Model Selection, Ensemble Methods (Boosting, Bagging, Random Forests)

3. Unit-IV Modeling Sequence/Time- Series Data, Deep Learning (Deep generative models, Deep Boltzmann Machines, Deep auto- encoders, Applications of Deep Networks) and Feature Representation Learning Unit- V Scalable Machine Learning (Online and Distributed Learning) Semi-supervised Learning, Active Learning, Reinforcement Learning, Inference in Graphical Models, Introduction to Bayesian Learning and Inference

4. Text Book(s): Kevin Murphy, Machine Learning: A Probabilistic Perspective, MIT Press, 2012. Trevor Hastie, Robert Tibshirani, Jerome Friedman, The Elements of Statistical Learning, Springer, 2017. Jiawei Han, Micheline Kamber, Jian Pei , Data Mining: Concepts and Techniques, 3/e, Morgan Kaufmann, 2016. Christopher Bishop, Pattern Recognition and Machine Learning, Springer, 2016. 19EEC334A: MACHINE 9 January 2023

5. 19EEC334A: MACHINE 9 January 2023

6. A(AI) Making machine to think, analyze and make decision

7. A(AI) Making machine to think, analyze and make decision

8. 19EEC334A: MACHINE 9 January 2023

9. Machine Learning (ML)

10. 9 January 2023 Department of EECE, GIT 19EEC334A: MACHINE LEARNING 21

11. 22

12. 23

13. 24

14. 25

15. A Typical Supervised Learning Workflow (for Classification)

16. A Typical Supervised Learning Workflow (for Classification)

17. A Typical Supervised Learning Workflow (for Classification)

18. A Typical Unsupervised Learning Workflow (for Clustering) 19EEC334A: MACHINE 9 January 2023

19. A Typical Unsupervised Learning Workflow (for Clustering)

20. A Typical Reinforcement Learning Workflow

21. 19EEC334A: MACHINE

22. 47

23. 48

24. 49

25. 50

26. 51

27. Department of EECE, GIT 19EEC334A: MACHINE 52

28. 53

29. 54

30. 55

31. 56

32. 57

33. 58

34. 59

35. 60

36. 61

37. 62

38. 63

39. 64

40. 65

41. 66

42. 67

43. 68

44. 69

45. 70

46. 71

47. 72

48. 73

49. 74

50. 75

51. 76

52. 77

53. 78

54. 79

55. 19EEC334A: MACHINE 9 January 2023 80

56. 19EEC334A: MACHINE 9 January 2023 81

57. 19EEC334A: MACHINE 9 January 2023 82

58. 19EEC334A: MACHINE 9 January 2023 83

59. 19EEC334A: MACHINE 9 January 2023 84

60. 19EEC334A: MACHINE 9 January 2023 85

61. 19EEC334A: MACHINE 9 January 2023 86

62. 19EEC334A: MACHINE 9 January 2023 87

63. 19EEC334A: MACHINE 9 January 2023 88

64. 19EEC334A: MACHINE 9 January 2023 89

65. 19EEC334A: MACHINE 9 January 2023 90

66. 19EEC334A: MACHINE 9 January 2023 91

67. 19EEC334A: MACHINE 9 January 2023 92

68. Linear Regression Line using least square method Goodness of Fit (Performance metric) R Square value MSE  ( yp  y ) 2  ( y  y ) 2  R 2 n 19EEC334A: MACHINE 9 January 2023  ( yp  y ) 2 MSE  On Board: Solve Example On Board: Solve R square Give assignment of MSE

69. 19EEC334A: MACHINE 9 January 2023

70. 19EEC334A: MACHINE 9 January 2023

71. 19EEC334A: MACHINE 9 January 2023

72. 19EEC334A: MACHINE 9 January 2023

73. 9 January 2023 INE 98 Y^= Predicted 19EEC334A: MACH Y= Actual Y^= Predicted Y= Actual

74. 19EEC334A: MACHINE 9 January 2023 99

75. 19EEC334A: MACHINE 9 January 2023 100

76. 19EEC334A: MACHINE 9 January 2023 101

77. 19EEC334A: MACHINE 9 January 2023 102

78. 19EEC334A: MACHINE 9 January 2023 103

79. 19EEC334A: MACHINE 9 January 2023 104

80. 19EEC334A: MACHINE 9 January 2023 106

81. 9 January 2023 Department of EECE, GIT 19EEC334A: MACHINE LEARNING 107 Y^= Predicted Y= Actual

82. Y^= Predicted 19EEC334A: MACHINE 9 January 2023 Y= Actual

83. 19EEC334A: MACHINE 9 January 2023

84. 19EEC334A: MACHINE 9 January 2023

85. 19EEC334A: MACHINE 9 January 2023

86. 19EEC334A: MACHINE 9 January 2023

87. 19EEC334A: MACHINE 9 January 2023

88. 19EEC334A: MACHINE 9 January 2023

89. 9 J anuary 2023 Department of EECE, GIT 19EEC334A: MACHINE LEARNING 115 Y^= Predicted Y= Actual

90. 9 January 2023 Department of EECE, GIT 19EEC334A: MACHINE 116

91. 9 January 2023 Department of EECE, GIT 19EEC334A: MACHINE 117

92. 19EEC334A: MACHINE 9 January 2023 118

93. 19EEC334A: MACHINE 9 January 2023 119

94. 19EEC334A: MACHINE 9 January 2023

95. Applications of Linear Regression 19EEC334A: MACHINE 9 January 2023 Some popular applications of linear regression are: Analyzing trends and sales estimates Salary forecasting Real estate prediction Arriving at ETAs in traffic.

96. 19EEC334A: MACHINE 9 January 2023

97. 19EEC334A: MACHINE 9 January 2023

98. 19EEC334A: MACHINE 9 January 2023

99. 19EEC334A: MACHINE 9 January 2023

100. 19EEC334A: MACHINE 9 January 2023

101. 9 January 2023 Department of EECE, GIT 19EEC334A: MA LEARNING Dividing data set into two subsets: Training set— a subset to train a model. Test set— a subset to test the trained model. CHINE

102. 19EEC334A: MACHINE 9 January 2023

103. 19EEC334A: MACHINE 9 January 2023

104. 19EEC334A: MACHINE 9 January 2023

105. 19EEC334A: MACHINE 9 January 2023

106. 9 January 2023 Department of EECE, GIT 19EEC334A: MACHINE LEARNING 132 The sigmoid function is a mathematical function used to map the predicted values to probabilities. It maps any real value into another value within a range of and 1. The value of the logistic regression must be between and 1, which cannot go beyond this limit, so it forms a curve like the "S" form. The S- form curve is called the Sigmoid function or the logistic function. In logistic regression, we use the concept of the threshold value, which defines the probability of either or 1. Such as values above the threshold value tends to 1, and a value below the threshold values tends to 0.

107. 9 January 2023 Department of EECE, GIT 19EEC334A: MACHINE 133

108. 19EEC334A: MACHINE 9 January 2023

109. The math behind Logistic regression 19EEC334A: MACHINE 9 January 2023

110. 19EEC334A: MACHINE 9 January 2023

111. 19EEC334A: MACHINE 9 January 2023

112. 19EEC334A: MACHINE 9 January 2023

113. 19EEC334A: MACHINE 9 January 2023

114. 19EEC334A: MACHINE 9 January 2023

115. 19EEC334A: MACHINE 9 January 2023

116. Type of Logistic Regression: On the basis of the categories, Logistic Regression can be classified into three types: Binomial: In binomial Logistic regression, there can be only two possible types of dependent variables, such as or 1, Pass or Fail, etc. Multinomial: In multinomial Logistic regression, there can be three or more possible unordered types of the dependent variable, such as "cat", "dogs", or "sheep“ Ordinal: In ordinal Logistic regression, there can be three or more possible ordered types of dependent variables, such as "low", "Medium", or "High". 19EEC334A: MACHINE 9 January 2023

117. 19EEC334A: MACHINE 9 January 2023

118. 19EEC334A: MACHINE 9 January 2023

119. 19EEC334A: MACHINE 9 January 2023

120. Y^= Predicted Y= Actual 19EEC334A: MACHINE 9 January 2023 Y^= Predicted Y= Actual

121. 19EEC334A: MACHINE 9 January 2023