Data analysis goals. Business Intelligence (BI) systems for business analysis Some benefits from using BI solutions

Small business in the CIS countries does not yet apply data analysis for business development, definitions of correlations, the search for hidden patterns: entrepreneurs are bypass by reports of marketers and accountants. The leaders of small and partially medium enterprises rely more on their intuition than the analysis. But at the same time, the analyst has a huge potential: it helps to reduce costs and increase profits, faster and objective to make decisions, optimize processes, better understand customers and improve the product.

Accountant will not replace analytics

The leaders of small enterprises often believe that the reports of marketers and accountants are fairly adequately reflecting the company's activities. But on the basis of dry statistics, it is very difficult to make a solution, and an error in counting without a profile formation is inevitable.

Case 1. Post-analysis of accommodation campaigns. By the new year, the entrepreneur declared a share in which certain goods were offered at a discount. Evaluating the revenue for the New Year's period, he saw how sales increased, and delighted his resourcefulness. But let's take into account all the factors:

Sales growing greatly on Friday, a day, when the maximum revenue is a weekly trend.
If you compare with increasing sales, which usually happens under the New Year, then the winnings are not so great.
If you filter out the exact goods, it turns out that sales indicators worsened.

Case 2. Study of turnovering. The store of women's clothing complexity with logistics: goods on some warehouses in deficiency, and on some lies for months. How to determine without an analysis of sales, how many trousers will start in one region, and how much coat send to another, while getting the maximum profit? To do this, you need to calculate the commodity, the ratio of sales speed and the average inventory for a certain period. If we put it easier, turning is an indicator of how many days the store will sell goods how quickly the average stock is sold, as the product pays off quickly. Store large reserves are economically unprofitable, as it freezes capital, slows development. If the reserve is reduced, a deficit may appear, and the company will again launch profits. Where to find a golden middle, the ratio in which the product is not stored in stock, and at the same time you can give a certain guarantee that the client will find the right unit in the store? For this, the analyst should help you determine:

desirable turnover
dynamics of turning.

When calculating with suppliers with a delay, you also need to calculate the ratio of credit line and turnover. Turnover in days \u003d average commodity supply * Number of days / trade over this period.

The calculation of the range of assortment and total shopping is helps to understand where it is necessary to move part of the goods. It is worth counting and what is the turnover of each unit of the range to make a grade solution under reduced demand, appreciation with elevated, moving to another warehouse. By categories you can develop a turnover report in this form. It can be seen that T-shirts and jumpers are sold faster, but a coat is long enough. Will this work be able to hold an ordinary accountant? We doubt. At the same time, the regular calculation of commodity and the application of results can increase profits by 8-10%

In which areas will the data analysis apply?

Sales. It is important to understand why sales go well (or bad), what is the dynamic. To solve this task, you need to explore the factors of influence on profit and revenue - for example, to analyze the length of the check and revenue to the buyer. Such factors can be explored by groups of goods, seasons, shops. You can identify elevations and sales pit, analyzing returns, cancellation and other operations.
Finance. Monitoring indicators need any financier to monitor the Kesflow and the distribution of assets on various fields of business. It helps to assess the effectiveness of taxation and other parameters.
Marketing. Any marketing company needs forecasts and post-analysis of shares. At the Study of the idea, it is necessary to identify groups of goods (control and target) for which we create a proposal. This is also a job for analytics of data, since the usual marketer does not have the necessary tools and skills for good analysis. For example, if for the control group, the amount of revenue and the number of buyers are equally greater than in comparison with the target - the action has not worked. To determine this, the interval analysis is needed.
Control. Have leadership qualities are not enough for the company's leader. Quantitative staff assessments in any case are needed for competent management of the enterprise. The effectiveness of managing the wage fund, the salary and sales ratio is important to understand the same way as the effectiveness of processes - for example, traffic accidents or movement employment during the day. It helps to properly distribute working time.
Web analysis. The site needs to competently, so that it becomes a sales channel, and this requires the right promotion strategy. Here will help you a web analysis. How to apply it? Study behavior, age, gender and other characteristics of customers, activity on certain pages, clicks, traffic channel, product performance, and so on. This will help improve business and website.
Assortment management. ABC analysis is extremely necessary to manage the assortment. The analyst must distribute the goods according to the characteristics to carry out such a type of analysis and understand which product is the most profitable, which is at the heart, and from what it is worth getting rid of. To understand the stability of sales, the XYZ analysis is well.
Logistics. More understanding about purchases, goods, their storage and accessibility will give the study of logistics indicators. The loss and needs of the goods, the commodity supply is also important to understand for successful business management.

These examples show how wide the ability to analyze data even for small enterprises. An experienced director will increase the company's profit and will benefit from the most minor information, using the data analysis correctly, and the work manager will significantly simplify visual reports.

The main goal of any data analysis is the search and detection of patterns in the amount of data. In business analysis, this goal becomes even wider. It is important for any supervisor not only to identify patterns, but also to find their cause. Knowledge of the reason will allow in the future to influence the business and makes it possible to predict the results of a particular action.

Data analysis targets for the company

If we talk about business, then the goal of each company win a competitive struggle. So the data analysis is your main advantage. It is he who will help you:

Reduce the cost of the company
Enlarge revenue
Reduce the time to fulfill business processes (learn a weak place and optimize it)
Increase the performance of the company's business processes
Perform any other goals aimed at improving the efficiency and effectiveness of the company's activities.

So, victory over competitors is in your hands. Do not rely on intuition. Analyze!

Data analysis goals for departments, departments, products

Oddly enough, but the targets listed above are fully suitable for analyzing the activities of departments, product analysis or advertising campaign.

The goal of any data analysis at any level is to identify regularity and take advantage of this knowledge to improve the quality of the product or the work of the company, the department.

Who needs data analysis?

All. Indeed, any company, from any sphere of activity, any department and any product!

In which areas can I use data analysis?

Production (construction, oil and gas, metallurgy, etc.)
Retail
Ecommerce.
Services
And many others

What departments can be analyzed within the company?

Accounting and Finance
Marketing
Advertising
Administration
Other.

Indeed, companies from any sphere, any departments within the company, any activities can, and it is important to analyze.

What can help BI analysis systems

BI-analysis systems, automated analytics systems, Big Data for analyzing large data are software solutions that already have built-in data processing functional, prepare them for analysis, analyzing itself and is the main thing - to visualize the analysis results.

Not every company has analysts department, or at least a developer who will serve the analytical system and databases. In this case, such Bi-analysis systems come to the rescue.

Today, more than 300 solutions are presented on the market. Our company focused on the Tableau solution:

In 2018, Tableau 69 times became the leader of Gartner's research among BI solutions
TableAu is easy to master (and our workshops confirm this)
For a full start of working with TableAu, the knowledge of the developer or statistics is not required

At the same time, companies that are already working with Tableau, say that to draw up reports that previously gathered in Excel for 6-8 hours, now takes no more than 15 minutes.

Do not believe? Try yourself - download the trial version of Tableau and get training materials for working with the program:

Download Tableau.

Download the full version of Tableau Desktop, 14 days and get as a gift Tableau business analyst training materials

About analyzing information has recently been said so much and so much that you can finally get confused in the problem. It is good that many pay attention to such an actual topic. It is only bad that under this term everyone understands what he needs, often without having a common picture on the problem. Fragmentality in this approach is the reason for the misunderstanding of what is happening and what to do. Everything consists of pieces, weakly interconnected and not having a common rod. Surely, you often heard the phrase "Patchwork Automation". With this problem, many have repeatedly encountered many and can confirm that the main problem with this approach is that it is almost never impossible to see the picture as a whole. With the analysis, the situation is similar.

In order for the place and purpose of each analysis mechanism, let's consider it entirely. It will be repelled from how a person makes decisions, since explaining how thought is born, we are not able to concentrate on how information technologies can use information technologies in this process. The first option is a decision-making person (LPR) uses a computer only as a means of extracting data, and the conclusions do it yourself. To solve this kind of tasks, reporting systems are used, multi-dimensional data analysis, diagrams and other imaging methods. The second option: the program not only retrieves the data, but also conducts various kinds of predicament, for example, cleaning, smoothing and so on. And the data processed in this way applies mathematical methods of analysis - clusterization, classification, regression, etc. In this case, the LPR receives not raw, but the data that has passed serious processing, i.e. Man is already working with models prepared by the computer.

Due to the fact that in the first case, almost everything that is connected actually with decision-making mechanisms is imposed on a person, the problem with the selection of an adequate model and the choice of processing methods is carried out beyond the analysis mechanisms, i.e., the base for decision making is either an instruction (for example How can I implement deviation response mechanisms), or intuition. In some cases, this is quite enough, but if the LPR is interested in knowledge that are deep enough, if you can express it, then simply data extraction mechanisms will not help. More serious treatment is needed. This is the very second case. All used predicability and analysis mechanisms allow LPD to work at a higher level. The first option is suitable for solving tactical and operational tasks, and the second is to replicate knowledge and solving strategic problems.

An ideal case would have the opportunity to apply both approaches to analysis. They allow you to cover almost all the needs of the organization in the analysis of business information. Variating techniques depending on the tasks, we will be able to squeeze the maximum of the available information anyway.

The overall work scheme is shown below.

Often, when describing a product analyzing business information, apply the type of risk management, forecasting, market segmentation ... But in reality, the solution of each of these tasks is reduced to the use of one of the analysis methods described below. For example, forecasting is the task of regression, market segmentation is clustering, risk management is a combination of clustering and classification, other methods are possible. Therefore, this technology set allows you to solve most business tasks. In fact, they are atomic (basic) elements from which a solution is collected by a task.

Now we describe each fragment of the scheme separately.

The primary data source must be the database of enterprise management systems, office documents, the Internet, because it is necessary to use all the information that can be useful for decision making. Moreover, it is not only internal for the organization of information, but also on external data (macroeconomic indicators, competitive environment, demographic data, etc.).

Although the data warehouse does not implement analysis technologies, it is the base on which the analytical system needs to be built. In the absence of data warehouse for collecting and systematizing the information necessary for the analysis of the information, most of the time will be leaving that it will largely reduces all the advantages of the analysis. After all, one of the key indicators of any analytical system is the ability to quickly get the result.

The next element of the scheme is the semantic layer. Regardless how the information will be analyzed, it is necessary that it is clear to the LPR, since in most cases the analyzed data is located in various databases, and the LPR should not delve into the nuances of the DBMS, then it is required to create a certain mechanism transforming the terms. The subject area in the calls of access mechanisms to the database. This task is performed by the semantic layer. It is desirable that it be one for all analysis applications, thus easier to apply various approaches to the task.

Reporting systems are intended to answer the question "What is happening". The first option is to use: regular reports are used to control the operational situation and analysis of deviations. For example, the system daily prepares reports on the balance of products in stock, and when its value is less than the average weekly sale, it is necessary to respond to this preparation for the purchase order, i.e., in most cases, this is standardized business operations. Most often, some elements of this approach in one form or another are implemented in companies (even if it is simply on paper), but it cannot be allowed that this is the only one of the available approaches to the data analysis. The second version of the application of reporting systems: processing of non-elected requests. When LPR wants to test any thought (hypothesis), he needs to get food for reflections confirming either a refuting idea, since these thoughts come spontaneously, and there is no accurate idea of \u200b\u200bwhat kind of information will need a tool that allows you to quickly and In a convenient view, this information get. The extracted data is usually submitted either as tables or in the form of graphs and diagrams, although other views are also possible.

Although various approaches can be used to build reporting systems, the most common today is an OLAP mechanism. The main idea is to present information in the form of multidimensional cubes, where the axes are measurements (for example, time, products, clients), and in cells are placed indicators (for example, sales amount, the average procurement price). The user manipulates measurements and receives information in the desired section.

Due to the simplicity of understanding OLAP, the data was widespread as a data analysis mechanism, but it is necessary to understand that its possibilities in the field of deeper analysis, for example, forecasting, are extremely limited. The main problem in solving, forecasting tasks is not the possibility of extracting the data of the data in the form of tables and diagrams, but the construction of an adequate model. Further everything is quite simple. A new information is applied to the input model, passed through it, and the result is the forecast. But the construction of the model is a completely nontrivial task. Of course, several ready-made and simple models can be laid into the system, for example, linear regression or something similar, quite often this is exactly what they do, but this problem does not decide. Real tasks almost always go beyond such simple models. And consequently, such a model will detect only obvious dependencies, the value of the detection of which is insignificant, which is so well known and so, or will build too gross forecasts, which is also completely uninteresting. For example, if you, when analyzing the stock exchange rate on the stock market, proceed from a simple assumption that tomorrow the stock will cost as much as today, then in 90% of cases you guess. And how valuable such knowledge is? Interest for brokers represent only the remaining 10%. Primitive models in most cases give the result of an approximately the same level.

The correct approach to building models is their step-by-step improvement. Starting from the first, relatively rough model, it is necessary as new data accumulates and the application of the model in practice improves it. Actually, the task of building forecasts and similar things go beyond the framework of the mechanisms of reporting systems, therefore it is not worth waiting in this direction of positive results when using OLAP. To solve a deeper analysis task, a completely different set of technologies are used, called Knowledge Discovery in Databases.

Knowledge Discovery in Databases (KDD) is the process of converting data into knowledge. KDD includes data preparation issues, choosing informative features, data cleaning, application methods of Data Mining (DM), post-processing data, interpretation of the results obtained. Data Mining is a detection process in the "raw" data of previously unknown, non-trivial, practically useful and accessible to the interpretation of knowledge necessary for decision-making in various fields of human activity.

The attractiveness of this approach lies in the fact that, regardless of the subject area, we use the same operations:

Extract data. In our case, this requires a semantic layer.
Clear data. Application for the analysis of "dirty" data can be completely reduced to no analysis mechanisms applicable in the future.
Transform data. Various analysis methods require data prepared in a special form. For example, somewhere in the quality of inputs can only be used digital information.
Conduct, actually, analysis - Data Mining.
Interpret the results obtained.

This process is repeated iteratively.

Data Mining, in turn, provides a solution of only 6 tasks - classification, clustering, regression, association, sequence and analysis of deviations.

It's all that needs to be done to automate the process of extracting knowledge. Further steps are already doing an expert, he is LPR.

Interpretation of computer processing results is assigned to humans. Simply different methods give various food for reflection. In the simplest case, these are tables and diagrams, and in a more complex - model and rules. To completely exclude the participation of a person is impossible, because A one or another result does not matter until it is applied to a specific subject area. However, it is possible to replicate knowledge. For example, LPR with the help of any method determined which indicators affect the creditworthiness of buyers, and presented it as a rule. The rule can be made to the system of issuing loans and thus significantly reduce credit risks by putting their estimate to flow. At the same time, from a person engaged in the actual discharge documents, a deep understanding of the causes of one or another output is not required. In fact, this is the transfer of methods, once applied in industry, in the field of knowledge management. The main idea is the transition from one-time and non-unified methods to conveyor.

All about what was mentioned above, only the names of the tasks. And to solve each of them, various techniques can be applied, ranging from classical statistical methods and ending with self-learning algorithms. Real business tasks are almost always one of the above methods or a combination thereof. Almost all tasks - forecasting, market segmentation, risk assessment, assessment of the effectiveness of advertising campaigns, the assessment of competitive advantages and many others are reduced to the above. Therefore, having available tool, the decisive list of tasks, you can say that you are willing to solve any task of business analysis.

If you drew attention, we have not mentioned what tool will be used for analysis, what technologies, because The tasks and methods of their solutions do not depend on the tool. This is just a description of the competent approach to the problem. You can use anything, it is important only to cover the entire list of tasks. In this case, we can say that there is a truly full-featured solution. Very often, mechanisms covering only a small part of tasks are proposed as a "full-featured solution of problems of business analysis". Most often, only OLAP is understood under the business analysis system, which is completely insufficient for full analysis. Under the thick layer of advertising slogans is only a reporting system. The spectacular descriptions of a particular analysis tool are hidden the essence, but it is enough to repel from the proposed scheme, and you will understand the actual position of things.

Business Analytics and Data Analysis. Effective consulting is what is needed for the qualitative development of any business. Resolving existing problems and crises, preventing potential, search for moves to increase profits and efficiency as a whole: All this provides you with high-quality consulting.

The consulting process is a complex, multistage, multi-level, there is no clear and universal approach to absolutely any case: business context, its niche, industry, ka, features and more: all this affects how business processes will be diagnosed. Naturally, the final stamp of consulting is preceded by many other pre-processes, such as the preparation of the task, description of business processes, business analytics, infrastructure diagnostics in general and IT infrastructure of the organization, in particular, analyzed the data, and already on the basis of this a number of recommendations are created. . It must be said that it is the business analyst and analysis of data that is the most important stages in the consulting process, it is they who lead to the appropriate conclusions, it is based on such an analysis, any recommendations are created.

Data analysis and business analyst: how to implement?

Qualitative analysis, in this case, can not do without any quantitative metrics. That is, it is very desirable that some automation of business processes, customer relations, suppliers, intermediaries, and all other processes have also been implemented in the work of the enterprise. It is with high-quality accounting for all processes occurring within the business, reporting and further analytics are greatly facilitated.

How can you automate document management, customer management and facilitate reporting?

The best option will be the exclusive software designed to perform a variety of tasks - from the FB Consult. You are offered high-quality customer management systems - various CRM genes intended for various business branches, an effective solution for monitoring document management - DocSvision, as well as software, suitable for business intelligence and data analysis, including - and to identify questionable financial transactions - QlikView . The introduction of such decisions will significantly increase the efficiency of your business.

Affordable work with Big Data using visual analytics

Improve business analytics and solve routine tasks using information hidden in Big Data using the TIBCO Spotfire platform. This is the only platform that provides an intuitive business users, a convenient user interface, which allows the entire spectrum of analytical technologies for large data without attracting IT professionals or the availability of special education.

The SpotFire interface allows the same convenient to work both with small data sets and multi-iteble clusters of large data: sensor readings, information from social networks, sales points or geolocation sources. Users with any levels of knowledge with ease work with informative control panels and analytical work processes simply using visualization, which are a graphical display of combining billions of data points.

Predictive analytics is training in the process of working on the basis of the company's joint experience for making more argued solutions. Using Spotfire Predictive Analytics, you can find new market trends from information received as a result of business analysts and take measures to minimize risks, which will improve the quality of management decisions.

Overview

Connecting to large data for high-performance analytics

Spotfire offers three main types of analytics with seamless integration with Hadoop and other major data sources:

On-Demand Analytics: Built-in user-based data connectors that simplify ultra-speed, interactive data visualization
Database Analysis (in-Database Analytics): Integration with distributional computing platform, which allows you to calculate data from any complexity based on large data.
RAM analysis (In-Memory Analytics): Integration with a statistical analysis platform, which takes data directly from any data source, including traditional and new data sources.

Together, all these integration methods represent a powerful combination of visual research and advanced analytics.
This allows business users to access, combine and analyze data from any data sources using powerful, convenient to use control panels and workflows.

Big Data Connectors

Spotfire Connectors for large data support all types of data access: in-datasource, in-Memory and on-Demand. Built-in Spotfire Data Connectors include:

Certified Hadoop Data Connectors for Apache Hive, Apache Spark SQL, Cloudera Hive, Cloudera Impala, DataBricks Cloud, HortonWorks, Mapr Drill and Pivotal Hawq
Other Certified Large Data Connectors include Teradata, Teradata ASTER and NETEZZA
Connectors for historical and current data from sources such as OSI PI sensors

In-Datasource Distributed Calculations

In addition to the convenient SPOTFIRE functionality of visual selection of operations for SQL queries, which contact data distributed in sources, SpotFire can create statistical and machine learning algorithms that operate inside data sources and return only the necessary results to create visualizations in the SpotFire system.

Users work with Dashboard with the functionality of the visual choice, which appeal to scripts using the built-in features of the TERR language,
Terr scripts are initiated by the work of distributed computing functional in collaboration with MAP / Reduce, H2O, SPARKR, OR FUZY LOGIX,
These applications, in turn, turn to high-efficiency systems as Hadoop or other data sources,
Terr can be deployed as an extended analytics engine at Hadoop nodes that are managed using MapReduce or Spark. TERR language can also be used for Teradata data nodes.
Results are visualized on Spotfire.

TERR for advanced analytics

TIBCO Enterprise Runtime for R (Terr) - Terr is a corporate level statistical package that was developed by Tibco for full compatibility with R language, implementing the company's long-term experience in an analytical system associated with S +. This allows customers to continue the development of applications and models not only using open R code, but also integrate and deployed its code R on a commercial reliable platform without having to rewrite your code. TERR has higher efficiency and reliable memory management, provides a higher data processing rate on large volumes in comparison with an open source R language.

Combining the whole functionality

Combining the aforementioned powerful functionality means that even in the case of the most complicated tasks that require analytics with a high level of reliability, users interact with simple and easy-to-use interactive work processes. This allows business users to visualize and analyze data, as well as share the results of analytics, without the need to know the details of the data architecture of the data underlying the business analysis.

Example: Spotfire interface for configuration, starting and visualizing the results of the model, which defines the characteristics of lost cargo. With this interface, business users can perform calculations using TERR and H2O (framework for distributed computing), referring to the transactions and shipments stored in the Hadoop clusters.

Analytical space for large data

Advanced and predictive analytics

Users use Spotfire Dreadboards with a visual selection functionality to launch a rich set of advanced features that make it easy to make forecasts, create models and optimize them during operation. Using large data, the analysis can be conducted within the data source (in-datasource), returning only aggregated information and the results required to create visualizations on the Spotfire platform.

Machine learning

A wide range of machine learning tools is available in the SpotFire built-in feature list, which can be used with one press. Statistics have access to the program code written in R language and can expand the functionality used. Functional machine learning can be divided with other users for easy reuse.

The following machine learning methods are available for continuous categorical variables on SpotFire and TERR:

Linear and logistic regression
Decision Trees, Random Forest Algorithm (Random Forest), Machine Busting (GBM)
Generalized linear (additive) models (Generalized AdDitive Models)
Neural networks

Content analysis

Spotfire provides analytics and visualization of data, a significant part of which was not used early - this is an unstructured text that is stored in sources such as documents, reports, the notes of CRM systems, website logs, publications on social networks and much more.

Located analytics

Multilayer high-resolution cards are an excellent way to visualize large data. A rich feature of SpotFire to work with cards allows you to create maps with so many reference and functional layers, which you need. Spotfire also makes it possible to use complex analytics while working with cards. In addition to geographical maps, the system creates cards to visualize user behavior, warehouses, production, raw materials and many other indicators.

Perhaps it will be useful to read: