Microsoft Azure Machine Learning

OK Introduction to MS Machine Learning
Review of “Microsoft Azure Machine Learning” by Sumit MundChapter 1:1. Starting from the initial pages, the prose is a bit verbose e.g. a. “Traditionally, it has been an area for experts. Developing and deploying a predictive modeling solution using machine learning has never been simple and easy, even for experts.” – OK, enough with the “experts” theme. We get it. It’s hard. One sentence would is enough to get the point across. This level of verbosity is a hallmark of the book as there are examples of this overstatement and restatement throughout. b. “You should not misunderstand that predictive analytics…” should be simply “You should understand…” the double negative is confusing and unnecessary. c. While not painfully obvious, it is clear that English is not the author’s native language. There are places where an inappropriate word or phrase is used e.g. “This kind of algorithm takes all of the data and groups them into different clusters…” – the use of the word “them” is unusual, a more common usage would be “…groups it into different clusters”. This is jarring to a native English speaking reader. d. In many places the word “the” is dropped inappropriately – again creating a jarring experience for one fluent in English. e. Phrases are sometimes mal-formed as if directly translated from a different language.2. Too much time is spent explaining basic themes such as defining project scope, the relative value and completeness of the data, etc. These things are already known to those of us in the software development field and for those that are professionals in their related engineering fields that would find this tool very useful, restating the obvious borders on insulting.3. Model reuse – again, too many examples. We get the notion of “code reuse”.4. Whomever the editor was for this book needs to be reviewed for competency : a. Missed an OBVIOUS FAIL when starting with the VERY FIRST EXAMPLE in the section entitled “Creating and running an experiment – do it yourself”. Item 1 on the list contains the following text: “…and then choose the Black Experiment …”. I’m pretty sure the author meant “BLANK” not “BLACK”. Oops. b. Incorrect word used: “A decision tree is a set of questions or decisions and their possible consequences arranged in a hierarchical FISSION” <- word wanted is “FASHION”. Oops. c. In the “No Free Lunch” section, the sentence that begins “So in practice” is as follows: “So in practice ,whenever” <- the command and space are transposed. Oops. The worst part is that Word points this out in glaring fashion.5. The example for uploading a data set describes how it’s done – not really useful as it’s just a matter of following the prompts. THERE IS NO EXAMPLE DATASET CREATED THAT WILL BE UPLOADED! What use is a process with no concrete application? Show me how but don’t actually walk me through the create data portion of the process and then upload it? Useless. A better way of doing this would have been to show the “Enter Data” portion, actually provide a few lines of data, save it to the PC, THEN use the upload process to fetch back the newly created data.OK, that’s the bad. I would say that approximately 1/5th of the book is verbose fluff/filler – it could be removed without detracting from the knowledge and intent of the book.In general, the book needs to be gone over by one who is actually fluent in English to correct the oddities that appear in the text and to fix the obvious grammar errors. That the first example was missed is a real fail, in my opinion.The Good:1. The book does a good job of introducing the user to the Azure Machine Learning portal. Myself, having never needed to use the engine, I think the author does a decent job of walking the reader through the necessary steps to get up and running.2. The author does not overburden the reader with too much information nor does the author pound a point to death – over speak it a bit, yes, but not to the point of frustration on the users part.3. The data sets chosen as examples vary widely giving the reader exposure to a large number of unrelated analysis tasks that should cover the areas of expertise of most readers, everything from flower analysis (the Iris example) to flight delays.4. The examples build one upon the other and, for the most part, follow a good, logical sequence (see above for the one suggestion for a change).5. The author even goes so far as to introduce customization via the “Python” and/or “R” languages without drowning the user in minutiae nor smothering them with syntax of the selected language.The summary:1. If I were searching for a book that would walk me through the basics of how to use the Microsoft Azure Machine Learning system, this would be a good book to use as a first guide.2. 4 stars out of 5.

Es una buena guía introductoria, actualizada y completa acerca del uso de ML Studio. Referente a la estadística, hace los señalamientos necesarios a las fórmulas pero se requiere de libros de apoyo o conocimientos sólidos.