In the year of the outbreak of artificial intelligence in 2017, what are the knowledge and skills for the value of machine learning, and the amount of gold? Let’s take a look!
First, the answer from Vladimir Novakovski:The biggest contribution to machine learning is usually the generalist. Especially in 2017, there are a lot of hype about machine learning. Many job seekers learn some in-depth courses online, which reminds me that in the 1990s, many people did not read computer science textbooks, but read some fast-track books called "20 days to learn VBScript." (In fact, there are still people like this today)
Still important skills include: (a) understanding the basics of statistics, optimization, and building quantitative models; and (b) understanding how models and data analysis are actually applied to products and businesses.
In addition to the above two points, the following skills are also crucial in 2017:
Know how to write high quality software. The time when one team writes quality junk software and the other team is perfect is over. Data and models can be easily processed using programming languages ​​such as Python and R and their software packages, so data scientists and machine learning engineers should be able to have a high level of programming skills and a basic understanding of system design.
Use large data sets. Although the term "big data" is used too frequently, the cost of data storage does show a sharp downward trend. This means that more and more data sets from different fields are processing and applying models.
Yes, once you have a basic understanding of the knowledge and the corresponding technical level, you will know at least one hot field, such as deep learning of computer vision and perception, recommendation engine, NLP (natural language processing), etc. benefit.
This is perhaps the most basic skill that data scientists must have—the work of data scientists is much more practical than the work of traditional statisticians. Programming is important in many ways, including the following three points:
Programming can enhance your ability to do statistics. If you have a lot of statistics, but there is no way to deal with them, then your statistical knowledge will be useless.
Programming gives you the ability to analyze large data sets. The dataset you work in the industry is not like the sample iris dataset (Iris dataset is a commonly used classification experimental dataset, collected by Fisher, 1936.) So cute, you can easily get millions or more The data.
By programming, you can create better data processing tools. This includes creating a visualization system for data, creating a framework for automated analysis experiments, and managing the company's data flow so that the required data can be reached.
Skill #2: Quantitative AnalysisQuantitative analysis is a must-have core skill for data scientists. Much of the science of data is about understanding the behavior of a particularly complex scientific system by analyzing the data produced by natural sciences and experiments. Quantitative analysis skills are important in many ways, including the following three points:
Experimental design and analysis: Especially for data scientists engaged in consumer Internet applications—the way data is recorded and the way the experiment works, providing a means for testing a large number of experimental hypotheses. Experimental analysis is very likely to go wrong (this can be asked any statistician), so data scientists can provide a lot of help in this regard.
Complex economy/growth system modeling: Some classic modeling is more common, such as customer churn models or customer lifetime value models. More complex modeling, such as supply and demand modeling, matching economically optimal methods between suppliers and suppliers, and modeling companies' growth channels to better quantify which growth paths are most valuable. The most famous example is Uber's price soaring model.
Machine Learning: Even if you don't implement a machine learning model, data scientists can also help create prototypes to test hypotheses, select and create functions, and determine the advantages and opportunities in existing machine learning systems.
Which data science people have the most demand for this skill? 1. Physicist 2. Statistician 3. Economist 4. Operations researcher 5. More, they are very accustomed to understanding complex systems (data inference) through top-down methods (models) or bottom-up approaches.
Skill #3: Product IntuitionProduct intuition is a skill that is related to the ability of data scientists to quantify the system. Product knowledge means understanding the complex systems that generate all the data that data scientists analyze. The importance of this skill is reflected in:
Hypothesis: A data scientist who knows a product very well can change the way the system behaves in a specific way. The assumption is based on the “premonition†about how certain aspects of the system behave, and you need to know how the system is premonitioning how it works.
Defining Metrics: Traditional analytical skill sets include identifying the primary and secondary metrics that a company can use to track the success of a particular target. Data scientists need to understand the product in order to create two product metrics: 1. Measure intent 2. Measure what has driving value.
Debug analysis: The "unbelievable" results are often caused by the "unbelievable" nature of the system. Good product knowledge helps speed product inspections and helps identify things that can go wrong faster.
Product knowledge typically involves the use of products created by the company. If that's not possible, then at least try to understand the people who actually use the product.
Skill #4: Communication skillsThis skill is important to help significantly increase the impact of all of these skills. This is especially important and is an important criterion for distinguishing between good data scientists and great data scientists. Good communication can be reflected in a variety of ways, including:
Communication insights: Some data scientists call it "storytelling." The most important thing here is to exchange insights in a clear, concise and effective way so that others in the company can effectively understand these insights.
Data visualization: A clear and clear chart is worth a thousand words.
Overall communication: Being a data scientist almost always means working as a team, including working with engineers, designers, product managers, operations staff, and more. Good overall communication helps promote trust and understanding, which is extremely important for those who are entrusted with managing data.
Skill #5: TeamworkThis last skill connects the above four skills. In particular, data scientists cannot exist in isolation and rely on teamwork. From what I have seen, when data scientists go deep into all aspects of the company (or at least in product development organizations), the results are best.
There are many reasons why teamwork is important, including:
Selfless: This includes helping and mentoring others and placing the company's mission on top of their personal career ambitions.
Constant iteration: Data scientists value feedback, and most of their work requires iterative and feedback with others to reach influential solutions.
Sharing knowledge with others: Since the data science profession is a newly emerging job, basically no one has complete skills, especially when you need to collect all the statistical techniques, frameworks, libraries, languages, and tools that may be useful. Because knowledge can be spread across different data scientists and their organizations, it is especially useful for data scientists to constantly share their knowledge, methods, and outcomes.
Stage Lights,Beam Lights,Stage Lights,Moving Head Lights
Guangzhou Chengwen Photoelectric Technology co.,ltd , https://www.cwstagelight.com