# Machine Learning Course

**Machine Learning Courses**

A brief summary of the topics covered in this course is as below. This is 150 hours course, it is suggested to complete this course in 2 Months. Apart from my classroom course, you will be given exercises, and it will take another 100 hours in the course duration to complete these exercises.

## Introduction to Machine Learning Supervised Learning

- What is Machine Learning?
- Supervised vs Unsupervised Learning
- Type of ML problems
- High Level view of ML Project Lifecycle

## Linear Regression

- Introduction to regression – equation, limitations
- Types of regressions
- Simple linear regression – Best-fit line, OLS, goodness of fit, Assumptions
- Model building
- Model Evaluation (regression parameters), Residual analysis and prediction, model interpretation
- The Mathematics of regression (parameter estimation using OLS, the gradient descent algorithm, ANOVA)
- Transformation of variables : Scaling and Standardization
- Polynomial regression
- Ordinary Least Squares
- Linear Regression
- Gradient Descent

## Multiple linear regression

- SLS vs MLR
- Multicollinearity
- Dummy Variable
- Polynomial regression
- Feature Selection
- Model Building: BACKWARD, FORWARD, STEPWISE
- R Square and Adjusted R Square
- Loss: RMSE , MSE, MAE Comparison
- Interpreting coefficients of MLR

## Regularization

- Introduction to Regularization
- Regularized linear models
- Ridge regression
- Lasso regression
- Elastic net

## Classification

- Introduction : Regression vs classification, types of classification, evaluating classification models
- Logistic Regression : Best-fit sigmoid curve, odds & log odds, multivariate logistic regression
- Building Logistic Regression Model
- Model Evaluation: Confusion metrics and accuracy, sensitivity & specificity, precision & recall, trade-offs, ROC-AUC, predictions
- Transformation of variables : Scaling and Standardization (optional)
- Decision Trees : Descriptive vs Discriminative classification, the decision tree algorithm, measuring purity (Gini index, Entropy, Information gain),
- Building Decision Trees Model
- K-Nearest Neighbor Model
**Telecom Churn Case Study**

## Ensamble Model

- Introduction to Ensemble Modelling
- Bagging (Bootstrap Aggregation) Model Introduction
- Random Forest
- Boosting Model Introduction
- Adaboost, Gradient Boost, XGBoost, Light GBM, CatBoost
- Stacking
- Bledning
- Out of Bag (OOB)
- Feature importance in random forests
- Building Random Forest Model
- Building Boost Based Model

## Support Vector Machine (SVM)

- Linear SVM classification
- Mathematical/ geometrical intuition
- In-depth geometrical intuition
- Soft margin classification
- Nonlinear SVM classification
- Polynomial kernel
- Gaussian, RBF kernel
- Data leakage
- SVM Regression
- Mathematical/ geometrical intuition

## Naïve Bayesian

- Introduction to Bayes theorem
- Multinomial naïve Bayes
- Gaussian naïve Bayes
- Various type of Bayes theorem and their intuition

## Clustering & Market Basket Analysis

- Introduction to clustering, types of clustering, Euclidean distance & centroid
- K-means clustering algorithm
- Transformation of variables : Scaling and Standardization (Optional)
- Building K-means model
- Introduction to market basket analysis, cross-selling & upselling, bag vs basket of products, the Apriori algorithm,
- Market Basket Analysis
- Gaussian Mixture Model
- K-Means
- K-Means++
- Batch K-Means
- Hierarchical Clustering
- DBSCAN
- Evaluation of clustering
- Homogeneity, completeness and v-measure
- Silhouette coefficient
- Davies-bouldin index
- Contingency matrix
- Confusion matrix

## Model Evaluation & Model Selection

- Principles of model selection – model & learning algorithm
- Simplicity, Complexity & overfitting, bias-variance trade off.
- Tuning Complexity and Regularization
- Regularization, hyperparameters, and cross validation
- Model building & Model evaluation
- Hyperparameter tuning using grid-search and randomized-search CV
- Handling class imbalance
- Model Selection

## Feature Engineering

- Feature engineering – introduction
- Handling numeric features, handling categorical features, handling time-based features
- Feature selection using CV
- Feature selection
- Recursive feature elimination
- Backward elimination
- Forward elimination
- Handling missing data
- Handling outliers
- Filter method
- Wrapper method
- Embedded methods
- Feature scaling
- Standardization
- Mean normalization
- Min-max scaling
- Unit vector
- Feature extraction
- PCA (principle component analysis)
- Introduction to Data encoding
- Nominal encoding
- One hot encoding
- One hot encoding with multiple categories
- Mean encoding
- Ordinal encoding
- Label encoding
- Target guided ordinal encoding
- Covariance
- Correlation check
- Pearson correlation coefficient
- Spearman’s rank correlation
- VIF

## Handling Imbalance Data

- Introduction to Data Imbalance
- Up-sampling
- Down-sampling
- Undersampling using Tomek Links
- K-Fold Cross Validation
- Stratified K-Fold
- Synthetic Minority Oversampling technique (SMOTE)
- Adjusting Class Weight
- Random Oversampling
- Data interpolation
- Choosing Right Evaluation Metric
- Treat problem as Anomaly Detection

## Model Evaluation Metrics

- Confusion Matrix
- Accuracy, Recall (Sensitivity/ TPR), Precision, F1, ROC, AUC
- Error Rate, Specificity, FPR, Prevalence
- RMSE, MAE, MSE
- R Square, Adjusted R Square

## Loss Function

- Introduction to Regression and Classification Loss Function
- Root Mean Square Error (RMSE)
- Mean Square Error (MSE)
- Mean Average Error (MAE)
- Huber Loss
- Maximum Likelihood Estimation
- Binary Cross Entropy Loss
- Hinge Loss
- Multi Class Cross Entropy Loss
- KL (Kullback Leibler) Divergence Loss

## Model Monitoring

- Introduction to model monitoring
- Model Drifting
- What to monitor?
- How frequently evaluate?
- How to take decision?

## Model Retraining

- Introduction to model retraining
- Retraining on same algorithm and new data
- Trying new features
- Trying new algorithms

## Dimensionality reduction

- The curse of dimensionality
- Dimensionality reduction technique
- PCA (principle component analysis) Introduction & Maths
- Scree plots
- Eigen-decomposition approach
- tNSE

## Decision Trees Based ML

- Decision Tree
- Definition of Ensemble techniques
- Bagging technique
- Bootstrap aggregation
- Random forest (bagging technique)
- Random forest repressor
- Random forest classifier
- Complete end-to-end project with deployment
- Adaboost, LGBM, XGBoost
- Gradient Boost

## Recommendation Systems

- Introduction to Recommendation Systems
- Application of Recommendation Systems
- Collaborative Filtering
- Content Based Filtering

## Multilayer Perceptron

## Hidden Markov Models (HMM)

## ML Libraries / Algorithm

- scipy (pandas, numpy, matplotlib, sympy, scikit-learn, scikit-image)
- scikit-learn, scikit-image, statsmodel