Random forest (Machine learning)

Random forest (Machine learning)

• 1.
  Random Forest Subject: MachineLearning Dr. Varun Kumar Subject: Machine LearningDr. Varun Kumar Lecture 8 1 / 13
• 2.
  Outlines 1 Introduction 2 WhyRandom Forest ? 3 What is Random Forest ? 4 Random Forest Example 5 How Random Forest Works ? 6 References Subject: Machine LearningDr. Varun Kumar Lecture 8 2 / 13
• 3.
  Introduction Supervised machine learning 1Regression Linear regression Logistic regression 2 Classification: It is process for dividing a data sets into a different categories or groups by adding label. Decision tree Naive Bayes Random forest K nearest neighbor (KNN) Subject: Machine LearningDr. Varun Kumar Lecture 8 3 / 13
• 4.
  Random Forest ⇒ Randomforest is an ensemble classifier made using decision tree models. ⇒ Ensemble model combines the results from different models. ⇒ Combination of multiple decision tree. Subject: Machine LearningDr. Varun Kumar Lecture 8 4 / 13
• 5.
  Continued– Subject: Machine LearningDr.Varun Kumar Lecture 8 5 / 13
• 6.
  Why random forest? Use case- Credit risk detection Subject: Machine LearningDr. Varun Kumar Lecture 8 6 / 13
• 7.
  Continued– 1 To minimizethe loss (Bank need a decision rule to predict for giving an approval to the loan.) 2 An applicant demographic (income, debit/credit history and socio-economic profiles are considered.) 3 Data science based assistance tool (Helps for modeling the behavioral patterns of individual customer) Variable Measurements Marital status Married or unmarried Gender Male or female Age Varried Status Default or not Time of payment Varried Employment Employed or un-employed Home ownership With home or without home Education level Secondary above or below Subject: Machine LearningDr. Varun Kumar Lecture 8 7 / 13
• 8.
  What is randomforest ? ⇒ Random forest is versatile algorithm and capable with Regression Classification ⇒ It is a type of ensemble learning method. ⇒ Commonly used predictive modeling and machine learning techniques. Subject: Machine LearningDr. Varun Kumar Lecture 8 8 / 13
• 9.
  Random forest algorithm T:Number of features D: Number of trees to be constructed Subject: Machine LearningDr. Varun Kumar Lecture 8 9 / 13
• 10.
  How random forestworks Days Outlook Humidity Wind Play 01 Sunny High Weak No 02 Sunny High Strong No 03 Overcast High Weak Yes 04 Rain High Weak Yes 05 Rain Normal Weak Yes 06 Rain Normal Strong No 07 Overcast Normal Strong Yes 08 Sunny High Weak No 09 Sunny Normal Weak Yes 10 Rain Normal Weak Yes 11 Sunny Normal Strong Yes 12 Overcast High Strong Yes 13 Overcast Normal Weak Yes 14 Rain High Strong ‘ No Subject: Machine LearningDr. Varun Kumar Lecture 8 10 / 13
• 11.
  Continued– Subject: Machine LearningDr.Varun Kumar Lecture 8 11 / 13
• 12.
  Features of randomforest 1 Most accurate learning algorithm 2 Works well for both classification and regression problem. 3 Runs efficiently on large data base 4 Requires almost no input preparation 5 Performs implicit features selection 6 Can be easily grown in parallel 7 Methods for balancing error in unbalanced data set. Important steps ⇒ Data acquisition ⇒ Divide data set → (1) Training data set (2) Testing data set ⇒ Implement model ⇒ Visualize ⇒ Model validation Subject: Machine LearningDr. Varun Kumar Lecture 8 12 / 13
• 13.
  References E. Alpaydin, Introductionto machine learning. MIT press, 2020. T. M. Mitchell, The discipline of machine learning. Carnegie Mellon University, School of Computer Science, Machine Learning , 2006, vol. 9. J. Grus, Data science from scratch: first principles with python. O’Reilly Media, 2019. Subject: Machine LearningDr. Varun Kumar Lecture 8 13 / 13