Scala Machine Learning Projects: Build real-world machine learning and deep learning projects with Scala (English Edition)

Cover
Copyright and Credits
Packt Upsell
Contributors
Table of Contents
Preface
Chapter 1: Analyzing Insurance Severity Claims
	Machine learning and learning workflow
		Typical machine learning workflow
	Hyperparameter tuning and cross-validation
	Analyzing and predicting insurance severity claims
		Motivation
		Description of the dataset
		Exploratory analysis of the dataset
		Data preprocessing
	LR for predicting insurance severity claims
		Developing insurance severity claims predictive model using LR
	GBT regressor for predicting insurance severity claims
	Boosting the performance using random forest regressor
		Random Forest for classification and regression
	Comparative analysis and model deployment
		Spark-based model deployment for large-scale dataset
	Summary
Chapter 2: Analyzing and Predicting Telecommunication Churn
	Why do we perform churn analysis, and how do we do it?
	Developing a churn analytics pipeline
		Description of the dataset
		Exploratory analysis and feature engineering
	LR for churn prediction
	SVM for churn prediction
	DTs for churn prediction
	Random Forest for churn prediction
	Selecting the best model for deployment
	Summary
Chapter 3: High Frequency Bitcoin Price Prediction from Historical and Live Data
	Bitcoin, cryptocurrency, and online trading
		State-of-the-art automated trading of Bitcoin
			Training
			Prediction
	High-level data pipeline of the prototype
	Historical and live-price data collection
		Historical data collection
		Transformation of historical data into a time series
			Assumptions and design choices
			Data preprocessing
		Real-time data through the Cryptocompare API
	Model training for prediction
	Scala Play web service
		Concurrency through Akka actors
		Web service workflow
			JobModule
			Scheduler
			SchedulerActor
			PredictionActor and the prediction step
			TraderActor
	Predicting prices and evaluating the model
	Demo prediction using Scala Play framework
		Why RESTful architecture?
		Project structure
		Running the Scala Play web app
	Summary
Chapter 4: Population-Scale Clustering and Ethnicity Prediction
	Population scale clustering and geographic ethnicity
		Machine learning for genetic variants
	1000 Genomes Projects dataset description
	Algorithms, tools, and techniques
		H2O and Sparkling water
		ADAM for large-scale genomics data processing
		Unsupervised machine learning
			Population genomics and clustering
		How does K-means work?
		DNNs for geographic ethnicity prediction
	Configuring programming environment
	Data pre-processing and feature engineering
		Model training and hyperparameter tuning
			Spark-based K-means for population-scale clustering
			Determining the number of optimal clusters
			Using H2O for ethnicity prediction
		Using random forest for ethnicity prediction
	Summary
Chapter 5: Topic Modeling - A Better Insight into Large-Scale Texts
	Topic modeling and text clustering
		How does LDA algorithm work?
	Topic modeling with Spark MLlib and Stanford NLP
		Implementation
			Step 1 - Creating a Spark session
			Step 2 - Creating vocabulary and tokens count to train the LDA after text pre-processing
			Step 3 - Instantiate the LDA model before training
			Step 4 - Set the NLP optimizer
			Step 5 - Training the LDA model
			Step 6 - Prepare the topics of interest
			Step 7 - Topic modelling 
			Step 8 - Measuring the likelihood of two documents
	Other topic models versus the scalability of LDA
	Deploying the trained LDA model
	Summary
Chapter 6: Developing Model-based Movie Recommendation Engines
	Recommendation system
		Collaborative filtering approaches
			Content-based filtering approaches
			Hybrid recommender systems
			Model-based collaborative filtering
		The utility matrix
	Spark-based movie recommendation systems
		Item-based collaborative filtering for movie similarity
			Step 1 - Importing necessary libraries and creating a Spark session
			Step 2 - Reading and parsing the dataset
			Step 3 - Computing similarity
			Step 4 - Testing the model
		Model-based recommendation with Spark
			Data exploration
			Movie recommendation using ALS
				Step 1 - Import packages, load, parse, and explore the movie and rating dataset
				Step 2 - Register both DataFrames as temp tables to make querying easier
				Step 3 - Explore and query for related statistics
				Step 4 - Prepare training and test rating data and check the counts
				Step 5 - Prepare the data for building the recommendation model using ALS
				Step 6 - Build an ALS user product matrix
				Step 7 - Making predictions
				Step 8 - Evaluating the model
	Selecting and deploying the best model 
	Summary
Chapter 7: Options Trading Using Q-learning and Scala Play Framework
	Reinforcement versus supervised and unsupervised learning
		Using RL
		Notation, policy, and utility in RL
			Policy
			Utility
	A simple Q-learning implementation
		Components of the Q-learning algorithm
			States and actions in QLearning
			The search space
			The policy and action-value
			QLearning model creation and training
		QLearning model validation
		Making predictions using the trained model
	Developing an options trading web app using Q-learning
		Problem description
		Implementating an options trading web application
			Creating an option property
			Creating an option model
			Putting it altogether
		Evaluating the model
		Wrapping up the options trading app as a Scala web app
			The backend
			The frontend
		Running and Deployment Instructions
		Model deployment
	Summary
Clients Chapter 8: Subscription Assessment for Bank Telemarketing using Deep Neural Networks
	Client subscription assessment through telemarketing
		Dataset description
		Installing and getting started with Apache Zeppelin
			Building from the source
			Starting and stopping Apache Zeppelin
			Creating notebooks
		Exploratory analysis of the dataset
			Label distribution
			Job distribution
			Marital distribution
			Education distribution
			Default distribution
			Housing distribution
			Loan distribution
			Contact distribution
			Month distribution
			Day distribution
			Previous outcome distribution
			Age feature
			Duration distribution
			Campaign distribution
			Pdays distribution
			Previous distribution
			emp_var_rate distributions
			cons_price_idx features
			cons_conf_idx distribution
			Euribor3m distribution
			nr_employed distribution
		Statistics of numeric features
		Implementing a client subscription assessment model
		Hyperparameter tuning and feature selection
			Number of hidden layers
			Number of neurons per hidden layer
			Activation functions
			Weight and bias initialization
			Regularization
	Summary
Chapter 9: Fraud Analytics Using Autoencoders and Anomaly Detection
	Outlier and anomaly detection
	Autoencoders and unsupervised learning
		Working principles of an autoencoder
		Efficient data representation with autoencoders
	Developing a fraud analytics model
		Description of the dataset and using linear models
		Problem description
		Preparing programming environment
			Step 1 - Loading required packages and libraries
			Step 2 - Creating a Spark session and importing implicits
			Step 3 - Loading and parsing input data
			Step 4 - Exploratory analysis of the input data
			Step 5 - Preparing the H2O DataFrame
			Step 6 - Unsupervised pre-training using autoencoder
			Step 7 - Dimensionality reduction with hidden layers
			Step 8 - Anomaly detection
			Step 9 - Pre-trained supervised model
			Step 10 - Model evaluation on the highly-imbalanced data
			Step 11 - Stopping the Spark session and H2O context
		Auxiliary classes and methods
	Hyperparameter tuning and feature selection
	Summary
Chapter 10: Human Activity Recognition using Recurrent Neural Networks
	Working with RNNs
		Contextual information and the architecture of RNNs
		RNN and the long-term dependency problem
		LSTM networks
	Human activity recognition using the LSTM model
		Dataset description
		Setting and configuring MXNet for Scala
	Implementing an LSTM model for HAR
		Step 1 - Importing necessary libraries and packages
		Step 2 - Creating MXNet context
		Step 3 - Loading and parsing the training and test set
		Step 4 - Exploratory analysis of the dataset
		Step 5 - Defining internal RNN structure and LSTM hyperparameters
		Step 6 - LSTM network construction
		Step 7 - Setting up an optimizer
		Step 8 - Training the LSTM network
		Step 9 - Evaluating the model
	Tuning LSTM hyperparameters and GRU
	Summary
Chapter 11: Image Classification using Convolutional Neural Networks
	Image classification and drawbacks of DNNs
	CNN architecture
		Convolutional operations
		Pooling layer and padding operations
			Subsampling operations
		Convolutional and subsampling operations in DL4j
			Configuring DL4j, ND4s, and ND4j
			Convolutional and subsampling operations in DL4j
	Large-scale image classification using CNN
		Problem description
		Description of the image dataset
		Workflow of the overall project
		Implementing CNNs for image classification
			Image processing
			Extracting image metadata
			Image feature extraction
			Preparing the ND4j dataset
			Training the CNNs and saving the trained models
			Evaluating the model
			Wrapping up by executing the main() method
	Tuning and optimizing CNN hyperparameters
	Summary
Other Books You May Enjoy
Index