Аналитика на все случаи жизни: Разновидность инструментов BI17 апреля 2014 Лидерам отраслей не нужно объяснять, почему чрезвычайно важно не только хранить информацию, а управлять ею и принимать взвешенные решения, основанные на качественном анализе данных. Сегодня, как никогда ранее, существует множество инструментов BI, но лишь некоторые из них оправдают ожидания пользователя. Чтобы выбрать правильное решение, предлагаем Вам ознакомиться с разновидностями современных решений бизнес-аналитики. (Материал опубликован на английском языке)
The ability for the enterprise to continue to make data-based decisions is becoming increasingly reliant on analytics. At present there are four principle categories of analytics, including:
- Descriptive: Descriptive Analytics are historically
based and utilize technologies found in conventional Business
Intelligence platforms such as reporting and dashboards.
- Diagnostic: Diagnostic Analytics are more interactive
than the descriptive variety, and commonly pertain to Data
Discovery tools and visualizations in which individuals can
layer data to determine trends based on data in real time or close
to real time.
- Predictive: Predictive Analytics calculate the
likelihood of any variety of future events while utilizing both
historic and recently acquired data.
- Prescriptive: Largely considered a variety of Predictive Analytics, Prescriptive Analytics go a step further and prescribe courses of action based on mutable algorithms that evolve as data is updated.
Most data-savvy enterprises incorporate Descriptive Analytics; more progressive ones will augment this approach with diagnostic capabilities. However, the full potential of Big Data, the Internet of Things, and of incorporating semi-structured and unstructured data is best realized with Predictive and Prescriptive Analytics, which are alternatively referred to as Advanced Analytics. The most recent Gartner Magic Quadrant for Business Intelligence notes that:
“By 2015, ‘smart data discovery,’ which includes natural-language query and search, automated, prescriptive advanced analytics and interactive data discovery capabilities, will be the most in-demand BI platform user experience paradigm, enabling mainstream business users to get insights (such as clusters, segments, predictions, outliers and anomalies) from data.”
Descriptive Analytics are the most basic form of analytics and are circumscribed in the fact that they are largely based on historic data which is typically structured. These analytics are associated with the traditional, IT-led centralized BI tools which can take a considerable amount of time to generate reports. Although these tools can enable ad-hoc queries and dashboard-style reports, their principle weakness is their reliance on historic data.
Descriptive Analytics are able to endure with contemporary relevance largely due to their application to social analytics and the explosion of social media. Social analytics merely tally and provide basic mathematical operations regarding data found on social media – basic tallying, multiplication, division and arithmetic. Due to their longevity and their relevance to social media, Descriptive Analytics are still the most widely deployed class of analytics in the enterprise today.
Diagnostic Analytics are most widely associated with Data Discovery tools, that class of BI software which enables users to analyze data in real time without lengthy, time-consuming reports. Interactive visualizations and dashboards allow users to easily manipulate data and layer different metrics atop each other to produce insights, while other discovery tools (search, data mashups, and in-memory analytics) enable laymen to effortlessly and expediently parse through data.
The most useful of the emerging applications of Diagnostic Analytics is geospatial or location intelligence. The algorithms for these analytics focus on time and space and are useful for calculating routes, mapping, and discerning distances based on a variety of mutable factors such as weather, customer location, and other consumer demographics. Location intelligence tools utilize three dimensional visualizations, clustering, and geographic data from a multitude of sources that allow users to layer data on interactive maps.
Predictive Analytics provide insight into the future of the most viable outcomes as indicated by a multitude of data types. These technologies include forecasting algorithms and are able to draw comparisons and relationships between variables with both recent and historic data to deliver the likelihood of what the user determines is a desirable – or undesirable – event. Specific tools for Predictive Analytics include data mining and modeling. In addition to presenting a scenario for the probability of an event occurring, Predictive Analytics can also provide the timeframe for when it might occur.
Data Modeling is at the core of Predictive Analytics, for the simple fact that once the proper model is created, individuals can determine the likelihood of future data from both recent and historical data. By taking the data that an enterprise already has, it is then able to hypothesize the direction of the trend of that data that should take place in the future. This capacity to predict future developments within a particular industry can be enough to give organizations a significant advantage over their competitors. There are a number of applications of Predictive Analytics that are beneficial to various aspects of the enterprise, including:
- Social Media: Social media applications for Predictive
Analytics include the ability to aggregate and gauge sentiment data
related to a particular service or product. This capability is
particularly useful for designing future products and services, or
altering them in a way so that customers positively respond to them.
Additionally, users can effectively predict the influence of a
particular person or entity’s social media presence, which is
valuable for targeting marketing efforts.
- Marketing: Other marketing gains derived from Predictive
Analytics pertain to specialization of product and sales pitches
based on existing customer behavior, which can influence promotions
and targeting. These analytics are also useful for advantageously
creating and adjusting pricing.
- Risk: In financial and insurance industries, Predictive
Analytics are deployed to determine customer behavior and important
factors such as risk assessment. Results include decreased incidence
of fraud and more efficient management of collateral and liquidity
- Operations/Human Resources: By analyzing traits of existing or previously successful employees, management can determine what sorts of individuals are likely to have similar or comparable traits that can assist in predicting their efficacy in a certain position.
Although the specific application that is most beneficial to a particular organization depends on its area of specialty, with the right modeling and requisite data the uses for Predictive Analytics are nearly endless.
Prescriptive Analytics go a step beyond Predictive Analytics and actually suggest a number of courses of actions (or prescriptions) that are based on the predictions of a certain outcome. Prescriptive Analytics are inherently predictive by nature, yet distinct in that they determine what the best course of action may be for any number of hypothetical situations that may occur. Like Predictive Analytics, Prescriptive Analytics readily incorporates both structured and unstructured data to present a more comprehensive view of the future.
One of the boons of Prescriptive Analytics is that they are extremely flexible and have the ability to improve with experience – much like cognitive computing. According to a recent article on Gigaom, “Prescriptive analytics algorithms recalibrate themselves. As the incoming data evolves so do the algorithms – they re-fit, re-predict and re-prescribe.” The implication is that organizations can not only plan for worst-case scenarios, but also analyze how to most effectively exploit positive ones.
There is a high degree of specificity in Prescriptive Analytics, which are applicable to specific domains and processes (for business or operations). Predictive Analytics have made successful forays in industries in which Big Data is the norm, such as fossil fuel industries.
Each of the aforementioned codifications of analytics – Descriptive, Diagnostic, Predictive, and Prescriptive – represents a successive advancement in sophistication and utility of the statistics involved. Descriptive Analytics can present an overview of the past, Diagnostic Analytics can inform the present, Predictive Analytics can inform the future, and Prescriptive Analytics can do so in a way that the enterprise can capitalize on it. Whereas most data-driven processes in organizations utilize the first two, the capabilities of Predictive Analytics can be applied to nearly every situation in which there is data to determine outcomes.
Ultimately, the efficacy of Big Data initiatives may well come to rest on the ability of the enterprise to incorporate predictive and Prescriptive Analytics. As indicated by Gartner’s first ever Magic Quadrant for Advanced Analytic Platforms:
“The rapid growth in available data, particularly new sources of data – such as unstructured data from customer interactions and streaming volumes of machine-generated data – require greater levels of sophistication from users and systems to be able to capture their full potential.”
Advanced Analytics provides that sophistication and can play a crucial difference in the success of competitors.