Chapter 1, Welcome to Machine Learning Using the .NET Framework, contextualizes machine learning in the .NET stack, introduces some of the libraries that we will use throughout the book, and provides a brief primer to F#.

Chapter 2, AdventureWorks Regression, introduces the business that we will use in this book—AdventureWorks Bicycle company. We will then look at a business problem where customers are dropping orders based on reviews of the product. It looks at creating a linear regression by hand, using Math.NET and Accord.NET to solve this business problem. It then adds this regression to the line of business application.

Chapter 3, More AdventureWorks Regression, looks at creating a multiple linear regression and a logistic regression to solve different business problems at AdventureWorks. It will look at different factors that affect bike sales and then categorize potential customers into potential sales or potential lost leads. It will then implement the models to help our website convert potential lost leads into potential sales.

Chapter 4, Traffic Stops – Barking Up the Wrong Tree?, takes a break from AdventureWorks. You will put on your data scientist hat, use an open dataset of traffic stops, and see if we can understand why some people get a verbal warning and why others get a ticket at a traffic stop. We will use basic summary statistics and decision trees to help in understanding the results.

Chapter 5, Time Out – Obtaining Data, stops with introducing datasets and machine learning models and concentrates on one of the hardest parts of machine learning—obtaining and cleaning the data. We will look at using F# type providers as a very powerful language feature that can vastly speed up this process of "data munging".

Chapter 6, AdventureWorks Redux – k-NN and Naïve Bayes Classifiers, goes back to AdventureWorks and looks at a business problem of how to improve cross sales. We will implement two popular machine learning classification models, k-NN and Naïve Bayes, to see which is better at solving this problem.

Chapter 7, Traffic Stops and Crash Locations – When Two Datasets Are Better Than One, returns back to the traffic stop data and adds in two other open datasets that can be used to improve the predictions and gain new insights. The chapter will introduce two common unsupervised machine learning techniques: k-means and PCA.

Chapter 8, Feature Selection and Optimization, takes another break from introducing new machine learning models and looks at another key part of building machine learning models—selecting the right data for the model, preparing the data for the model, and introducing some common techniques to deal with outliers and other data abnormalities.

Chapter 9, AdventureWorks Production – Neural Networks, goes back to AdventureWorks and looks at how to improve bike production by using a popular machine learning technique called neural networks.

Chapter 10, Big Data and IoT, wraps up by looking at a more recent problem—how to build machine learning models on top of data that is characterized by massive volume, variability, and velocity. We will then look at how IoT devices can generate this big data and how to deploy machine learning models onto these devices so that they become self-learning.

You will need Visual Studio 2013 (any version) or beyond installed on your computer. You can also use VS Code or Mono Develop. The examples in this book use Visual Studio 2015 Update 1.

The lines between business computing and scientific computing are becoming increasingly blurred. Indeed, an argument can be made that the distinction was never really as clear as it has been made out to be in the past. With that, machine learning principles and models are making their way into mainstream computing applications. Consider the Uber app that shows how far Uber drivers are from you, and product recommendations built into online retail sites such as Jet.

Also, the nature of the .NET software developer's job is changing. Earlier, when the cliché of ours is a changing industry was being thrown around, it was about languages (need to know JavaScript, C#, and TSql) and frameworks (Angular, MVC, WPF, and EF). Now, the cliché means that the software developer needs to know how to make sure their code is correct (test-driven development), how to get their code off of their machine onto the customer's machine (DevOps), and how to make their applications smarter (machine learning).

Also, the same forces that are pushing the business developer to retool are pushing the research analyst into unfamiliar territory. Earlier, analysts focused on data collection, exploration, and visualization in the context of an application (Excel, PowerBI, and SAS) for point-in-time analysis. The analyst would start with a question, grab some data, build some models, and then present the findings. Any kind of continuous analysis was done via report writing or just re-running the models. Today, analysts are being asked to sift through massive amounts of data (IoT telemetry, user exhaust, and NoSQL data lakes), where the questions may not be known beforehand. Also, once models are created, they are pushed into production applications where they are continually being re-trained in real time. No longer just a decision aid for humans, research is being done by computers to impact users immediately.

The newly-minted data scientist title is at the confluence of these forces. Typically, no one person can be an expert on both sides of the divide, so the data scientist is a bit of a jack of all trades, master of none who knows machine learning a little bit better than all of the other software engineers on the team and knows software engineering a little bit better than any researcher on the team. The goal of this book is to help move from either software engineer or business analyst to data scientist.

In this book, you will find a number of text styles that distinguish between different kinds of information. Here are some examples of these styles and an explanation of their meaning.

Code words in text, database table names, folder names, filenames, file extensions, pathnames, dummy URLs, user input, and Twitter handles are shown as follows: "The Script1.fsx file is then added to the project."

A block of code is set as follows:

let multipliedAndIsEven = 
    ints
    |> Array.map (fun i -> multiplyByTwo i)
    |> Array.map (fun i -> isEven i)

Any command-line input or output is written as follows:

val multipliedAndIsEven : string [] =
  [|"even"; "even"; "even"; "even"; "even"; "even"|]

New terms and important words are shown in bold. Words that you see on the screen, for example, in menus or dialog boxes, appear in the text like this: "When the Add New Item dialog box appears, select Script File."

Feedback from our readers is always welcome. Let us know what you think about this book—what you liked or disliked. Reader feedback is important for us as it helps us develop titles that you will really get the most out of.

To send us general feedback, simply e-mail <[email protected]>, and mention the book's title in the subject of your message.

If there is a topic that you have expertise in and you are interested in either writing or contributing to a book, see our author guide at www.packtpub.com/authors.

Now that you are the proud owner of a Packt book, we have a number of things to help you to get the most from your purchase.