RANDOM FOREST for machine and deep learning for computer science

RANDOM FOREST for machine and deep learning for computer science

• 1.
  RANDOM FOREST GONZAGA EDWARDOKANE 2019/HD05/25243U EGOR ELEAZAR 2019/HD05/25247U
• 2.
  AREAS OF DISCUSSION •What is a decision tree and how does it work • Key terms • How does a decision tree work • What is Random Forest • Why use Random Forest • How Random Forest works • Applications of random Forest • Use Case
• 3.
  WHAT IS ADECISION TREE AND HOW DOES IT WORK? • Is a tree shaped diagram used to determine a course of action. • Each branch of the tree represents a possible decision, occurrence, or reaction.
• 4.
  KEY TERMS 1. Entropy– is the measure of randomness or unpredictability in the dataset
• 5.
  2. Information gain– the measure of decrease in entropy after the dataset is split
• 6.
  3. Leaf Node– carries the classification or the decision 4. Decision Node – has two or more branches
• 7.
  5. Root Node– The top most decision node
• 8.
  HOW DOES ADECICION TREE WORK Problem statement: To classify the different types of fruits in the bowl based on different features • The dataset (Bowl) is quite messy and has a high entropy • To split the data, we have to frame the conditions that split the data in such a way that the information gain is the highest NB: Gain is the measure of decrease in entropy after splitting
• 9.
  • We willtry to choose a condition that gives us the highest gain • We will do that by splitting the data using each condition and checking the gain that we get out of them NB: The condition that gives us the highest gain will be used to make the first split
• 10.
  • After splittingbased on diameter, the entropy has reduced
• 11.
  • We thensplit the right node further based on color • We can then predict a lemon with 100% accuracy • Apple can also be predicted with 100% accuracy
• 12.
  WHAT IS RANDOMFOREST • Is a method that operates by constructing multiple decision trees • Bunch of decision trees bundled together • Based on the idea “The wisdom of the crowd” • Gets predictions from each tree and selects the best solution by voting • The decision of majority of the trees is chosen by the random forest as the final decision • Example - Getting recommendations from friends for vacation destinations • Can be used for regression and classification
• 14.
  WHY DO WEUSE RANDOM FOREST? 1. No overfitting • In overfitting, the model learns “too much” from the training data set • Overfitting is the case where the overall cost is really small, but the generalization of the model is unreliable. • What use is a model that has learned very well from the training data but still can’t make reliable predictions for new inputs? • We always want to find the trend, not fit the line to all the data points • Training time is less
• 15.
  2. High Accuracy •Random forest runs efficiently on large databases • Produces highly accurate predictions for large data 3. Estimates missing data • Maintains accuracy when a large proportion of data is missing • E.g. different sets of demographic statistics coming in from various areas where;  One set is missing number of children in the house  Another set missing size of the house Random forest will look at the sets differently and build 2 different trees, then guesses which one fits better
• 16.
  HOW RANDOM FORESTWORKS • Step 1: Select the random samples from a given dataset • Step 2 : Construct a decision tree from each sample and get a prediction result from each decision tree • Step 3: Perform a vote for each predicted result • Step 4: Select the prediction result with the most votes as the final prediction
• 17.
  • Lets takethis blackened fruit and try to classify it • This is an example where random forest works really good when missing data Diameter = 3 Colour = Orange Grows in summer = Yes Shape = Circle
• 19.
  APPLICATIONS 1. Kinect • Gameconsole developed by Microsoft. • Uses infrared to track body movements and recreates it in the game.
• 20.
  2. Remote Sensing •Used in Enhanced Thematic Devices (ETM) on satellites to acquire high-resolution imaging information of the Earth’s surface • Less training time • Higher accuracy
• 21.
  3. Object Detection •Multiclass object detection e.g. Traffic where the algorithm is used in sorting out different types of vehicles such as buses, lorries, etc • Provides better detection in complicated environments