Monday, April 17, 2017

Analyst and Data Scientist skill set in Digital Marketing. Applying tools and technologies to be successful. Part 2.



The business need in the analysis of data will only grow. In 2011, International Data Corporation estimated the weight of the world's data in 1.8 zettabytes (1.8 trillion GB), two years later, in 2013, they were already 4.4 zettabytes. Last year, only aggregate Internet traffic exceeded 1.1 zettabytes. Daily more than 2.5 quintillions of data bytes are generated, and for a year their volume is at least doubled.
A competent digital analytic is required today for every company working on the Internet, including marketing field. The demand for data scientists is also growing. According to IDC calculations, US companies already by 2018 will need 181 thousand specialists in the analysis of big data. It is a rapidly growing professional sphere, where tools and technologies are changing rapidly too. There are several main entry points in the data science for digital analysts: Excel, SQL, SAS, R, and Python.
Excel
Pretty simple and familiar to everyone. The set of solved analytical problems is limited only by the knowledge, skills and imagination.
SQL
SQL is universal and therefore popular with analysts. It is used both inside SAS, and in R, and in Python. Allows to create databases, select data from tables according to specified conditions and easily group them.
"Godfather" analysts: used in business since 1976. It is easier to learn than R and Python. It has detailed documentation and customer support, so many companies that have been on SAS for a long time rarely go over to something else. But younger organizations often use R or Python.
The SAS certificate is an advantage for job seekers. But training is expensive, and the SAS system is costly for companies. To be in demand in the labor market, digital analytics is clearly worth learning R or Python.
R
This language has been used in development since 2007. It was used primarily in scientific and applied research, but quickly gained momentum in business. The popularity of R continues to grow, defying the almost 40-year-old monopoly of SAS. Analysts appreciate this language for its simplicity and functionality, including for free tools. For R there are about 12 thousand statistical, graphic and analytical packages.
Thanks to the open Python code, it's a great free alternative to R. Its advantages are in code readability and availability to the production environment. Many systems for producing large amounts of data use Python. There are also disadvantages: for example, there are more statistical packages in R. But for Python there are many libraries and modules that help the analyst to turn into Data Scientist.
Also there are many different sets of libraries and statistical programs which in different combinations give freedom in choosing options for working with data and creating forecasts:
In the basis of analytics for big data, you need SQL and Excel. To move further within the profession, it is worth choosing one of the languages depending on your needs - R, Python or SAS. And ideally - use all three. For example:
 • SAS - to process big data sets;
 • Python - create a model of machine learning;
 • R or SAS Visual Analytics- provide a graphical representation of the data.
The most important tools that an analyst needs are an understanding of what and why to calculate, and the ability to clearly express thoughts. Therefore, the skill to ask questions and correctly formulate the results will save hours of data processing. The main thing to remember: as in any case, 80% of the result is achieved due to the correct formulation of the problem and 20% of known heuristics. 



Analyst and Data Scientist skill set in Digital Marketing. Reasons to go to Web Analytics. Part 1.



Today most of the business somehow works in the network, and the main weapon of the online business is the data. And the more data, the more correct decisions can be made by company employees.
It is important to work with the data, because today it is the source of accurate answers to questions about the product and marketing. Previously, companies had to either guess, or rely on long-term studies.
At the beginning of the century, there was very little available data. Then companies learned how to work with analytics in real time, and the Internet marketer cope with the amount of information. Now there is a lot of data. So much that the specialist has to deal with it. Even a small online store gets 200-300 advertising campaigns a week. These campaigns are divided into different segments, different value propositions and go through several steps. To understand which campaigns work, companies need to track the path for each of them from the first touch to the target action. There are other sources, there is user behavior after purchase. In short, there is a lot of interesting.
The work with data looks like this: for example, a marketer needs to calculate ROI from the spent advertising campaigns. He asks the web analytics to build funnels for campaigns. Further, knowing the budget and the cost of production, the marketer concludes whether the campaign worked or not. Also product managers, business developers, traffic managers, even PR specialists need web analytics help. The web analyst is their eyes, without which the work gets complicated.
Data need to be collected, cleared and based on it correct conclusions can be made.
Collect - means to be able to find data. It's like hunting: you need to place traps in the right places, if you want someone to be caught.
Clean - sometimes extra data makes you only confused. The web analyst must know exactly what is important for making a decision. This requires an understanding of business metrics and business objectives.
Interpret - to make the right conclusion, you need to understand the logic of decision making. Ultimately, marketers interpret the data, but without understanding the logic, the web analyst risks giving the wrong answer.
Web-analysts should have following qualities: love of numbers, love for accuracy, meticulousness, curiosity, ingenuity. Students from the faculties of applied mathematics, statistics, sociology, and others, where there is an emphasis on numbers can try to work in this field, and just those who like to dig into the information.
5 reasons to look in the direction of web analytics.
1. The growing demand for specialists
A few years ago, only one digital company out of 20 had a web analyst. Today there is a web analyst in 1 out of 10 companies. Tomorrow this position will be in each company.
2. Career growth
Web-analyst works in tandem with other specialists and quickly absorbs knowledge from related professions. This helps to get horizontal growth. If you get bored, there is always marketing, business development and other professions, in which knowledge of web analytics systems also add weight.
3. Interesting problems
There are no two identical funnels. Web analytics is a constant explosion of the brain, a constant search for truth, digging deep into the information noise.
4. Self- assessment
Everyone wants to be friends with a web analyst, because spoiling the relationship it's easy to lose "vision."
5. Non-career development
Working with data is a useful skill in life. A person who devotes himself to this profession, perceives reality differently, becomes more pragmatic and more rational.

How about this news which proofs that data scientist and analyst are perspective professions. HR Company Glassdoor named 25 best jobs in the US for 2017, the leader of the rating is the data scientist profession. In the 2015 ranking, the data researcher's job was ninth, with the assistant doctor on the first place. The list also includes the profession of analyst, UX designer, hiring manager, tax manager, mobile application developer, marketing manager and electrical engineer. When compiling the rating, Glassdoor specialists were based on anonymous feedback from employees.



To learn more about this topic, check following blogs:

Avinash Kaushik. Web analytics career guide: from zero to hero in five steps!


Sergio Aicardi. 5 Reasons you can’t put off hiring a web analyst any longer.

Lizzie Plaugic. Radiohead’s saddest song is True Love Waits, according to data analysis. http://www.theverge.com/2017/2/22/14698446/radiohead-most-depressing-song-data-spotify-gloom-index

Wednesday, April 12, 2017

Status and future of Big Data. Part 3.

1970: the emergence of the concept of big data
The concept of "big data" arose in the time of mainframes and related scientific computer calculations. Knowledge-intensive computing has always been difficult and usually linked with the need to process big amounts of information.
2008: The appearance of the term "Big Data"
It is one of the few names that have a very reliable date of the birth - September 3, 2008, when a special issue of the oldest British scientific journal Nature was published, dedicated to finding the answer to the question “How can technologies that open up the possibility of working with big amounts of data influence future science?” The special issue summed up the previous discussions about the role of data in science in general and in electronic science (e-science) in particular. Along with the increase in processing power and the development of storage technologies, the possibilities of analyzing big data are gradually becoming available to small and medium-sized businesses and cease to be exclusively the prerogative of large companies and research centers. This contributes to the development of the cloud computing model.
In 2012 participants in the project "Internet and American Life" Pew Research Center's and "Imagining the Internet" center at the University of Elon asked the players of the digital field to evaluate two scenarios for the development of events by 2020.
53% of respondents agreed with the following forecast:
Big Data 2020 - thanks to many changes, including the creation of the "Internet of Things", by 2020 human and machine analysis of large amounts of data will improve the quality of public, political and economic research. The development of the phenomenon called "Big Data" will create the possibility of the emergence of "science casting" (broadcast events in real time), the development of "deductive software", analyzing the data structure in order to predict possible results, and creating algorithms for advanced correlations that give a new understanding of the world. In general, the presence of Big Data is a benefit for society.
39% agreed with the second statement:
Big Data 2020 - thanks to many changes, including the creation of the "Internet of things", by 2020 human and machine analysis of large amounts of data will create more problems than it will solve. Having a big amount of data will give a false confidence in our ability to predict the future, which will lead many people to serious and painful mistakes. Moreover, power-endowed people and institutions will use the analysis of Big Data to achieve their selfish goals, manipulate research results to justify their actions. The Big Data is also harmful, because it serves the interests of the majority (sometimes not quite accurately), belittling the importance of the minority and ignoring important outsiders. In general, the arrival of the Big Data is a disadvantage for the society.
Respondents could not choose both scenarios. A significant number of participants emphasize that, although they chose either a positive or a negative scenario, the real situation in 2020 will combine the features of both.
2015: Gartner excluded "Big Data" from popular trends
October 6, 2015, it became known about the exclusion from the Gartner report "Hype Cycle for Emerging Technologies 2015" information about big data. Researchers have explained this by the blurring of the term - the technologies included in the concept of "big data" have become the daily reality of business. The company explained its decision by the fact that the concept of "big data" includes a large number of technologies that are actively used in enterprises, they are partly related to other popular areas and trends and have become a daily working tool.



 Gartner "Hype Cycle for Emerging Technologies 2015"

The topic of big data has not disappeared, but has been transformed into many different scenarios:
Examples include precision farming, systems for countering fraud, systems in medicine, which allow diagnosing and treating patients at a qualitatively new level.
One of the main trends now is the Internet of things, which allows to connect machines with each other (machine-to-machine).

In May 2015, Andrew White, vice president of research at Gartner, reflected in his blog: "The Internet of Things will overshadow the big data, as too focused technology, it can generate several more effective solutions and tools, but it will be the Internet of things that will become the platform of the future that will increase our productivity in the long term."

Similar ideas were earlier - according to the results of the Gartner report for 2014, published Forbes columnist Gil Press: "We live in an era when it is important not only to be able to accumulate information, but to extract business benefits from it. The first were industry, which directly work with the consumer: telecommunications and banking, retail".

The reality of using big data, according to the Cisco Connected World Technology Report, is that most companies collect, record and analyze data. Nevertheless, the report says, many companies in connection with Big Data face a number of complex business and information technology problems.

Companies collect and use data of a wide variety of types, both structured and unstructured. Here are the sources from which the Cisco Connected World Technology Report data:
• 74% collect current data.
• 55% collect historical data.
• 48% take data from monitors and sensors.
• 40% use the data in real time, and then erase them. Most often, real-time data are used in India (62%), USA (60%) and Argentina (58%).

EMC: Big Data forecast and real-time analytics will be combined
In the nearest future we will get acquainted with a new chapter in the history of development of "big data" analytics as the two-level processing model develops. The first level will be a "traditional" Big Data analysis, where big data sets are not analyzed in real time. The new, second level will provide the ability to analyze relatively large volumes of data in real time, mainly through in-memory analytics. In this new phase of Big Data development, technologies such as DSSD, Apache Spark and GemFire ​​will be as important as Hadoop. The second level will offer us at the same time new and habitual ways of using "data lakes" - for "analytics on the fly" with the aim of influencing events, while they occur. This opens up new business opportunities

You can find more article here: http://haneenalansari.blogspot.com/2017/04/what-is-future-of-digital-marketing.html

Tuesday, April 11, 2017

Status and future of Big Data in an industry. Part 2-The difference between Business Intelligence and Big Data. Methods for analyzing Big Data.


The difference between Business Intelligence and Big Data

Craig Bati, executive director of marketing and technology director for Fujitsu Australia, said that business analysis is a descriptive process for analyzing the results achieved by the business in a certain period of time, while the processing speed of big data makes it possible to make predictive analysis capable of offering advice to businesses future. Big data technologies also allow to analyze more types of data in comparison with business intelligence tools, which makes it possible to focus not only on structured repositories.

Mac Slocum of O'Reilly Radar believes that although big data and business intelligence have the same goal (seeking answers to a question), they differ from each other in three aspects:

• Big data is designed to handle more significant amounts of information than business intelligence.
• Big data is designed to handle more quickly received and changing information, which means deep research and interactivity.
• Big data is intended for processing unstructured data (emails, videos, audio files, web pages, and social media messages)

According to Oracle's white paper, “Oracle Information Architecture: An Architect's Guide to Big Data”, when dealing with big data, we approach information in a different way than doing business analysis. When working with big data, the result is obtained during the process of data cleaning by sequential modeling: firstly, a hypothesis is put forward; secondly, a statistical, visual or semantic model is constructed; next, on the basis of this, the correctness of the hypothesis is checked and the following is advanced. This process requires the researcher to either interpret visual values ​​or compose interactive queries based on knowledge, or develop adaptive algorithms of “machine learning” that can obtain the desired result. And the lifetime of such an algorithm can be quite short.


Methods for analyzing Big Data

There are many different methods of analyzing data sets, which are based on tools borrowed from statistics and informatics (machine learning). The following methods reflect the most popular approaches. Some of the techniques do not necessarily apply only to big data and can be used for smaller arrays (A/B testing, regression analysis).
·         A/B testing.
·         Association rule learning.
·         Classification.
·         Cluster analysis.
·         Crowdsourcing.
·         Data fusion and data integration.      
·         Data mining.
·         Ensemble learning.
·         Genetic algorithms.
·         Machine learning.
·         Natural language processing (NLP).
·         Network analysis.
·         Optimization.
·         Pattern recognition.
·         Predictive modeling.
·         Regression.
·         Sentiment analysis.
·         Statistics.
·         Simulation.
·         Time series analysis.
·         Unsupervised learning (close to Cluster Analysis).


For more articles about Business Analytics you can check this blog: 

http://haneenalansari.blogspot.com/2017/04/how-can-you-get-great-job-in-business.html