Big Data technologies are now used everywhere - in industry, medicine, business, entertainment. So, without the analysis of big data, large retailers will not be able to work normally, sales at Amazon will fall, meteorologists will not be able to predict the weather for many days, weeks and months in advance. It is logical that big data specialists are now in great demand, and demand is constantly growing.
GeekBrains prepares representatives of this field, trying to provide students with both theoretical knowledge and teach by example, for which experienced experts are involved. This year Big Data analysts from the online university GeekUniversity and Russia's largest retailer X5 Retail Group have become partners. The company's specialists, having extensive knowledge and experience, helped to create a branded course, whose students receive both theoretical training and practical experience during the training.
We spoke with Valery Babushkin, director of modeling and data analysis at X5 Retail Group. He is one of data scientists in the world (30th in the world ranking of machine learning specialists). Together with other teachers, Valery tells GeekBrains students about A/B testing, mathematical statistics on which these methods are based, as well as modern practices for calculations and features of implementing A/B testing in offline retail.
Why do we need A/B tests at all?
This is one of the best methods for finding the best ways to improve conversions, economics, and behavioral factors. There are other ways, but they are more expensive and complicated. The main advantages of A/B tests are their relatively low price and availability for businesses of any size.
About A / B tests, we can say that this is one of the most important ways to find and make decisions in business, decisions on which both profit and development of various products of any company depend. Tests provide an opportunity to make decisions based not only on theories and hypotheses, but also on practical knowledge about how specific changes modify the interaction of clients with the network.
It is important to remember that in retail you need to test everything - marketing campaigns, SMS mailings, tests of the mailing lists themselves, the location of products on the shelves and the shelves themselves in the trading floors. If we talk about an online store, then here you can test the location of elements, design, inscriptions and texts.
A/B tests are a tool that helps a company, for example, a retailer, to always be competitive, sense changes in time and change itself. This allows the business to be as efficient as possible, bringing profit to the maximum.
What are the nuances of these methods?
The main thing is that there should be a goal or a problem on which testing will be based. For example, the problem is a small number of customers at a retail outlet or online store. The goal is to increase the influx of buyers. The hypothesis is that if the product cards in the online store are made larger, and the photos are brighter, then there will be more purchases. Next, an A / B test is carried out, the result of which is an assessment of the changes. After the results of all tests are received, you can begin to form an action plan for changing the site.
It is not recommended to conduct tests with overlapping processes, otherwise the results will be more difficult to evaluate. It is recommended that tests be carried out first on the highest priority goals and formulated hypotheses.
The test must last long enough for the results to be valid. How much depends, of course, on the test itself. So, on New Year's Eve, the traffic of most online stores increases. If before that the design of the online store was changed, then a short-term test will show that everything is fine, the changes are successful, and traffic is growing. But no, because no matter what you do before the holidays, traffic will grow, the test cannot be completed before the New Year or immediately after it, it must be long enough to reveal all the correlations.
The importance of the right relationship between the goal and the indicator being measured. For example, by changing the design of the same site of an online store, the company sees an increase in the number of visitors or customers and is satisfied with this. But in fact, the size of the average check may be less than usual, so the total income will be even lower. This, of course, cannot be called a positive result. The problem is that the company did not simultaneously check the link between the increase in visitors - the increase in the number of purchases - the dynamics of the size of the average check.
Is testing only for online stores?
Not at all. A popular method in offline retail is to implement a full pipeline to test hypotheses offline. This is the construction of a process in which the risks of incorrect selection of groups for the experiment are reduced, the optimal ratio of the number of stores, pilot time and the size of the estimated effect is selected. It is also the reuse and continuous improvement of post-effects analysis methodologies. The method is needed to reduce the likelihood of false acceptance errors and missing the effect, as well as to increase sensitivity, because even a small effect on the scale of a large business is of great importance. Therefore, you need to be able to identify even the slightest changes, minimize risks, including incorrect conclusions about the results of the experiment.
Retail, Big Data and real cases
Last year, X5 Retail Group specialists assessed the dynamics of sales volumes of the most popular products among 2018 World Cup fans. There were no surprises, but the statistics still turned out to be interesting.
So, βbestseller No. 1β was water. In the cities that hosted the World Cup, water sales increased by about 46%, Sochi turned out to be the leader, where the turnover increased by 87%. On match days, the highest figure was recorded in Saransk, where sales increased by 160% compared to normal days.
In addition to water, fans bought beer. From June 14 to July 15 in those cities where matches were held, the turnover of beer increased by an average of 31,8%. Sochi also became the leader - here beer was bought 64% more actively. But in St. Petersburg, the growth was small - only 5,6%. On the days of matches in the same Saransk, the volume of beer sales increased by 128%.
Research has also been carried out on other products. Data obtained on peak days of food consumption allows more accurate forecasting of demand in the future, taking into account event factors. An accurate forecast makes it possible to foresee the expectations of buyers.
During the testing of X5 Retail Group, two methods were used:
Bayesian structural models of time series with cumulative difference estimation;
Regression analysis with estimation of error distribution bias before and during the championship.
What else does retail use from Big Data?
- There are quite a lot of methods and technologies, from what can be called offhand, these are:
- Demand forecast;
- Optimization of the assortment matrix;
- Computer vision to detect voids on the shelves and detect the forming queue;
- Promo forecast.
Lack of specialists
The demand for experts in the field of Big Data is constantly growing. So, in 2018, the number of vacancies related to big data increased by 7 times compared to 2015. In the first half of 2019, the demand for specialists exceeded 65% of the demand for the whole of 2018.
Large companies especially need the services of Big Data analysts. For example, in Mail.ru Group they are needed in any project that processes text data, multimedia content, performs speech synthesis and analysis (this is, first of all, cloud services, social networks, games, etc.). The number of vacancies in the company has tripled over the past two years. In the first eight months of this year, Mail.ru hired as many Big Data specialists as in the entire past year. At Ozon, the Data Science department has tripled in the past two years. Megafon has a similar situation - the data analysis team has grown several times over the past 2,5 years.
Without a doubt, in the future, the demand for representatives of specialties related to Big Data will grow even stronger. So if there is an interest in this area, you should try your hand.
Source: habr.com