Duration
5 days
Overview
This module introduces to trainees on the fundamental concept of Big Data and lay the foundation of the participants with knowledge and skillset to implement basic big data algorithm including Regression, Classification and Clustering. At the end of the training, the participants shall be able to solve most big data problem using commonly used algorithm like Linear Regression, Decision Tree, Random Forest, KNN, K-Means and others. Participants shall also be equipped with knowledge to fine tuning the model, and evaluate to determine accuracy of the model.
On Day 5, we shall also learn on image processing including advance Deep Learning algorithm like Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNN).
Pre-requisite
Development experience
Target Audience
MINDEF developers
DAY 1:
This module introduces to trainees on the fundamental concept of Big Data and lay the foundation that prepare trainees for the rest of the training modules in the course. The first module in Big Data solution is the Data Ingestion, and this training shall focus on data sourcing, data cleansing and data manipulation.
Description
1Introduction to Speedminer
Getting Start with Speedminer.
1.1Login and User Account
1.2Folder and security
1.3Change Password
1.4Drop Down Menu, Slide Out Menu and Right Click Menu
1.5Logout
2Data Loading
Learn Interactive ETL tool to perform Data Ingestion.
2.1Loading Text File and Excel File
2.2Join
2.3Column Transformation
2.4Row Transformation
2.5Output into class
2.6Output into text file
2.7Hands-On Exercise
3Data Loading II
Speedminer comes with sample data for testing, beside sample data, there are other source of data from internet as well as database that we shall explore.
3.1Loading Sample Data
3.2Web Crawler
3.3Social Media – Load and Analyse
3.4Loading from Database
3.5Data Exploration
3.6Correlation check
3.6Hands-On Exercise
4Hands-On Exercise
Learn to develop use case and solve challenging problem
DAY 2:
As linear regression represents the simplest and easiest to understand machine learning algorithm, we shall also touch base this algorithm to give a complete picture to trainees on the concept behind Big Data.
Description
1Data Query
Produce query on data loaded
1.1Create a Query
1.2Save and Load
1.3Aggregation and Filter
1.4Hands-On Exercise
2API
To connect to external API to load information
2.1Connect to City Weather API
2.2Connect to Geolocation API
2.3Expose data through JSON
2.4Hands-On Exercise
3Machine Learning – Regression
Linear regression is one of the easiest and most popular Machine Learning algorithms. It is a statistical method that is used for predictive analysis
3.1Loading of Property Data
3.2Creation of Model
3.3Testing Data
3.4Model Evaluation
3.5Hands-On Exercise
3.6Features Selection
3.7Ordinary Least Squares
3.8Backward Elimination
3.9Testing Data
3.10Model Evaluation
3.11Hands-On Exercise
DAY 3
Classification in machine learning and statistics is a supervised learning approach in which the computer program learns from the data given to it and categorize the data into classes. This training shall equip participants with the knowledge and skillset to develop big data solution for classification problem.
Description
1Classification Concept
This chapter explain the concept of Classification and type of problem solved by classification machine learning algorithm.
1.1Understand Classification
1.2Binary Classification
1.3Multi-Class Classification
1.4Multi-Label Classification
1.5Imbalanced Classification
2Logistic Regression
Logistic regression is one of the most popular Machine Learning algorithms, which comes under the Supervised Learning technique. It is used for predicting the categorical dependent variable using a given set of independent variables.
2.1Sigmoid Function
2.2Assumptions for Logistic Regression
2.3Logistic Regression Equation
2.4Type of Logistic Regression
2.4.1Binomial
2.4.2Multinomial
2.4.3Ordinal
2.5Developing a Logistics Regression
2.5.1Data Preparation
2.5.2Training the Model
2.5.3Prediction
2.5.4Test Accuracy
2.5.5Visualization of Result
2.6Hands-on and exercises
3k-Nearest Neighbors (k-NN)
k-Nearest Neighbors algorithm (or k-NN for short) is a non-parametric method used for classification and regression. In both cases, the input consists of the k closest training examples in the feature space
3.1Understand k-NN
3.2Euclidean distance
3.3Select the value of k
3.4Pros and Cons of k-NN
3.5Developing a k-NN
3.5.1Data Preparation
3.5.2Training the Model
3.5.3Prediction
3.5.4Test Accuracy
3.5.5Confusion Matrix
3.5.6Visualization of Result
3.6Hands-on and exercises
4Support Vector Machine (SVM)
Support Vector Machine is one of the most popular Supervised Learning algorithms, which is used for Classification as well as Regression problems. However, primarily, it is used for Classification problems in Machine Learning. The goal of the SVM algorithm is to create the best line or decision boundary that can segregate n-dimensional space into classes so that we can easily put the new data point in the correct category in the future. This best decision boundary is called a hyperplane.
4.1Understand SVM
4.2Types of SVM
4.2.1Linear SVM
4.2.2Non-linear SVM
4.3Hyperplane and Support Vectors
4.4Developing an SVM
4.4.1Data Preparation
4.4.2Training the Model
4.4.3Prediction
4.4.4Test Accuracy
4.4.5Confusion Matrix
4.4.6Visualization of Result
4.5Hands-on and exercises
5Decision Tree
Decision tree learning uses a decision tree to go from observations about an item (represented in the branches) to conclusions about the item's target value (represented in the leaves).
5.1Understand Decision Support
5.2Terminology
5.3Attribute Selection Measures
5.3.1Information Gain (measurement of changes in Entropy)
5.3.2Gini Index
5.4Pruning: Getting an Optimal Decision tree
5.4.1Cost Complexity Pruning
5.4.2Reduced Error Pruning
5.5Pros and Cons of Decision tree
5.6Developing a Decision Support
5.6.1Data Preparation
5.6.2Training the Model
5.6.3Prediction
5.6.4Test Accuracy
5.6.5Confusion Matrix
5.6.6Visualization of Result
5.7Hands-on and exercises
6Random Forest
Random forests or random decision forests are Machine Learning method for classification, regression and other tasks, that operate by constructing a multitude of decision trees at training time and outputting the class that is the mode of the classes (classification) or mean prediction (regression) of the individual trees.
7.1Understand Random Forest
7.2Assumptions for Random Forest
7.3Applications of Random Forest
7.4Pros and Cons of Random Forest
7.5Developing a Random Forest
7.5.1Data Preparation
7.5.2Training the Model
7.5.3Prediction
7.5.4Test Accuracy
7.5.5Confusion Matrix
7.5.6Visualization of Result
7.6Hands-on and exercises
DAY 4
Clustering or cluster analysis is a machine learning technique, which groups the unlabelled dataset. It is a way of grouping data points into different clusters, consisting of similar data points. The objects with the possible similarities remain in a group that has less or no similarities with another group. This training shall equip participants with the knowledge and skillset to develop big data solution to group data into clusters.
Description
1Clustering Concept
This chapter explain the concept of Clustering and type of problem solved by clustering machine learning algorithm.
1.1Understand Classification
1.2Types of Clustering Methods
1.2.1Partitioning Clustering
1.2.2Density-Based Clustering
1.2.3Distribution Model-Based Clustering
1.2.4Hierarchical Clustering
1.2.5Fuzzy Clustering
1.3Clustering Algorithms
1.3.1K-Means algorithm
1.3.2Mean-shift algorithm
1.3.3DBSCAN Algorithm
1.3.4Expectation-Maximization Clustering using GMM
1.3.5Agglomerative Hierarchical algorithm
1.3.6Divisive Hierarchical algorithm
1.3.7Affinity Propagation
1.4Applications of Clustering
2K-Mean
K-means clustering aims to partition n observations into k clusters in which each observation belongs to the cluster with the nearest mean, serving as a prototype of the cluster
2.1Understand K-Mean
2.2Choose the value of "K number of clusters"
2.2.1Elbow Method
2.3Developing a Random Forest
2.3.1Data Preparation
2.3.2Finding the optimal number of clusters using the elbow method
2.3.3Training the Model
2.3.4Visualization of the Clusters
2.4Hands-on and exercises
3Mean-Shift Clustering
Meanshift is falling under the category of a clustering algorithm in contrast of Unsupervised learning that assigns the data points to the clusters iteratively by shifting points towards the mode (mode is the highest density of data points in the region, in the context of the Meanshift). As such, it is also known as the Mode-seeking algorithm. Mean-shift algorithm has applications in the field of image processing and computer vision
3.1Understand Mean-Shift Clustering
3.2Kernel Density Estimation
3.3Developing a Mean-Shift Clustering
3.3.1Data Preparation
3.3.2Training the Model
3.3.3Visualization - Surface plot and Contour plot
3.4Hands-on and exercises
4DBSCAN Algorithm
The full name of the DBSCAN algorithm is Density-based Spatial Clustering of Applications with Noise. It is an unsupervised learning algorithm for clustering. DBSCAN clustering is capable in creating clusters of arbitrary shapes and form clusters based on varying densities.
4.1Reachability and Connectivity
4.1.1Directly Density-Reachable
4.1.2Density-Reachable
4.1.3Density-Connected
4.2Parameter of DBSCAN Clustering
4.2.1Epsilon
4.2.2Neighbourhood
4.2.3min_sample
4.3Classification of data points
4.3.1Core points
4.3.2Boundary points
4.3.3Noise points
4.4Density edge and density connected points
4.4.1Density edge
4.4.2Density connected points
4.5How to determine epsilon and z
4.6Noise elimination
4.7Developing a DBSCAN algorithm
4.7.1Data Preparation
4.7.2Training the Model
4.7.3Visualization
4.8Hands-on and exercises
DAY 5
Convolutional Neural Networks (CNNs) leverage spatial information, and they are therefore well suited
for classifying images. In this module, we shall start from basic of image processing using KNN to the advance neural network (Deep Learning).
Description
1Classic Image Classification
Image classification is the task of assigning a label to an image from a predefined set of categories. Image classification is to analyse an input image and return a label that categorizes the image. The label is always from a predefined set of possible categories. This chapter we shall explore on Classic Image classification.
1.1Image Classification Algorithms
1.1.1Logistic Regression
1.1.2Support Vector Machines (SVMs)
1.1.3Random Forests
1.1.4KNN
1.1.5Artificial Neural Networks (ANN)
2.1Loading MNIST sample dataset
2.2Data manipulation and viewing of image
2.3Data Partition
2.4Image recognition using KNN classifier
2.5Developing an image classification using KNN classifier
2.5.1Data Preparation
2.5.2Training the Model
2.5.3Perform Prediction
2.5.4Model Evaluation
2.5.5Visualization
2.6Developing an image classification using Random Forest Classifier
2.6.1Data Preparation
2.6.2Training the Model
2.6.3Perform Prediction
2.6.4Model Evaluation
2.6.5Visualization
3Artificial Neural Network with Multilayer Perceptron
A multilayer perceptron (MLP) is a class of feedforward artificial neural network (ANN).
... An MLP consists of at least three layers of nodes: an input layer, a hidden layer and an output layer. Except for the input nodes, each node is a neuron that uses a nonlinear activation function.
3.1History of Neural Networks
3.2Multilayer Perceptron concept
3.3Deep Learning
3.4 Transfer Learning
4Stochastic Gradient Descent (SGD)
Stochastic gradient descent is a very popular and common algorithm used in various Machine Learning algorithms, most importantly forms the basis of Neural Networks.
4.1Understand Stochastic Gradient Descent
4.2Model evaluation
4.3Perform prediction
4.4Hands-on and exercises
5Convolutional Neural Networks
In deep learning, a convolutional neural network (CNN, or ConvNet) is a class of deep neural network, most commonly applied to analyze visual imagery.[1] They are also known as shift invariant or space invariant artificial neural networks (SIANN), based on the shared-weight architecture of the convolution kernels or filters that slide along input features and provide translation equivariant responses known as feature maps
5.1Convolution Filters
5.2Hierarchical Features
5.3Deep Learning and Transfer Learning
5.4Activation Functions
5.5Pooling and Fully Connected Layer (FC Layer)
5.6Softmax Classification technique
Programme scope, delivery arrangements and suitability are confirmed during enquiry.