دانلود کتاب یک راهنمای دوستانه برای علوم داده

Data science has been celebrated as the sexiest job of the twenty-first century because it has so much potential to help organizations understand themselves and their functions better through insights from data, enabling them to do whatever it is they do, just better. It’s easy to get excited about these possibilities, as many organization leaders have been doing. However, setting aside the fact that it’s probably a little early to be making pronouncements about an entire century that’s not even a quarter over, the two little words “data science” don’t really convey the amount of planning and work that goes into getting those helpful insights. It’s common, especially among nontechnical leaders (the majority in the corporate world), to hear the buzz terms “data science,” “machine learning,” and “AI” and think they will solve all of their problems. Often, these people will spin up a data science team—or sometimes just hire one data scientist—and expect the insights to start pouring out. More often than not, the hapless data scientists they’ve hired will discover that there’s insufficient data to do data science, especially good data science. Even the most skilled data scientist can’t turn water into wine—the mantra “garbage in/garbage out” is 100% true in the data world. This is unfortunate and often means that the work simply can’t get done. But one even more dangerous thing that can happen in the face of garbage in/garbage out is that inexperienced data scientists will produce garbage results—things that look insightful but are simply wrong. They may pass these faux insights along, and leaders may use them to make completely wrong-headed business decisions. An awareness of the requirements and limitations of data science is crucial to avoid this nightmare scenario and get meaningful insights. Understanding data science requirements and limitations also serves as a reminder of the time–quality–cost triad in all technical work. You can’t get high-quality data insights fast for cheap. At least one of the three will suffer when the other two are prioritized. There are always a lot of moving parts in a data science project. This book is intended to help you understand why that is, by showing the incredible breadth of topics data science involves. These include statistics, data analysis, programming, ethics, data security, data privacy, data engineering, machine learning (ML), natural language processing (NLP), data visualization, big data, cloud computing, and project management. Most projects only involve some of these areas, but it can be difficult to know far in advance exactly which ones will be relevant. So having a real appreciation for everything that should be considered when doing data science will help temper your expectations to more realistic levels. The last part of the book focuses on practical aspects of what it takes to become a data scientist. It follows that being an expert in all the above areas is unnecessary to be a good data scientist. It is still important to know what areas data science touches and the areas it operates in, because sometimes an expert in one of those other areas may be needed. Or perhaps you need to develop those skills yourself. Data science is a constantly changing field, and working in it means you never stop learning.