Chapter 1, Predict the Class of a Flower from the Iris Dataset, focuses on building a machine learning model leveraging a time-tested statistical method based on regression. The chapter draws the reader into data processing, all the way to training and testing a relatively simple machine learning model.
Chapter 2, Build a Breast Cancer Prognosis Pipeline with the Power of Spark and Scala, taps into a publicly available breast cancer dataset. It evaluates various feature selection algorithms, transforms data, and builds a classification model.
Chapter 3, Stock Price Predictions, says that stock price prediction can be an impossible task. In this chapter, we take a new approach. Accordingly, we build and train a neural network model with training data to solve the apparently intractable problem of stock price prediction. A data pipeline, with Spark at its core, distributes training of the model across multiple machines in a cluster. A real-life dataset is fed into the pipeline. Training data goes through preprocessing and normalization steps before a model is trained to fit the data. We may also provide a means to visualize the results of our prediction and evaluate our model after training.
Chapter 4, Building a Spam Classification Pipeline, informs the reader that the overarching learning objective of this chapter is to implement a spam filtering data analysis pipeline. We will rely on the Spark ML library's machine learning APIs and its supporting libraries to build a spam classification pipeline.
Chapter 5, Build a Fraud Detection System, applies machine learning techniques and algorithms to build a practical ML pipeline that helps find questionable charges on consumers’ credit cards. The data is drawn from a publicly accessible Consumer Complaints Database. The chapter demonstrates the tools contained in Spark ML for building, evaluating, and tuning a pipeline. Feature extraction is one function served by Spark ML that is covered here.
Chapter 6, Build Flights Performance Prediction Model, makes us able to leverage flight departure and arrival data to predict for the user if their flight is delayed or canceled. Here, we will build a decisions trees-based model to derive useful predictors, such as what time of the day is best to have a seat on a flight, with a minimum chance of delay.
Chapter 7, Building a Recommendation Engine, draws the reader into the implementation of a scalable recommendations engine. The collaborative-filtering approach is laid out as the reader walks through a phased recommendations-generating process based on users’ past preferences.
