Udacity Meeting (San Jose)
Winner Algorithm (XGBoost)
Pourya Ayria
Udacity Meeting (03/03/2018)
OutLine:
-
How XGBoost Works
-
Multi CLassification(Iris dataset)
-
Problem Definition
-
Load the Dataset
-
Analyze Data
3.1. Descriptive Statistics
3.2. Data Prepration: Reduce the size of data, Label Encoding
3.3. Data Visualization
3.4. Checking Target Distribution
-
Modeling
4.1. Feature Importance
4.2. Kfold Cross Validation
4.3. XGBoost vs. Logistic Regression
4.4 Tuning: GridSearchCV, RandomizedSearchCV
4.5 Prediction
-
Saving the Model