Shulou(Shulou.com)06/03 Report--

The conclusions interpreted by different data analysts with the same data may be different, or even completely opposite, but the conclusion itself is not right or wrong, so from objective data to subjective people, it is necessary to have some scientific analysis methods as bridges to help the information of the data be transmitted better, more comprehensively and faster. So, what are the common data analysis methods for products? Today, we will discuss it with you through the Rubik's Cube big data.

Product data analysis method

I. trend analysis

Trend analysis is generally used for long-term tracking of core indicators, such as click-through rate, GMV, number of active users. Generally make a simple data trend chart, but just making a data trend chart is not an analysis, we also need to observe what trend changes in the data, whether there is periodicity, whether there is an inflection point, and analyze the reasons behind it. whether it's internal or external.

The best output of trend analysis is the ratio. There are month-on-month comparison, year-on-year, fixed base ratio. For example, how much GDP increased in April 2017 than in March, this is month-on-month, which reflects the recent trend, but has a seasonal impact. In order to eliminate the seasonal effects, the year-on-year is launched, for example, how much GDP increased in April 2017 compared with April 2016, which is the same period last year. It is easier to understand the fixed base ratio, which is to fix a certain basis point, such as using the January 2017 data as the base point, while the fixed base ratio compares the May 2017 data with the January 2017 data.

II. Comparative analysis

Horizontal comparison: horizontal comparison is compared with yourself. The most common data indicator is to compare with the target value to answer whether we have achieved our goal or not, and to answer how much we have increased around the north compared with last month.

Vertical comparison: to put it simply, it is a comparison with others. We have to compare with our competitors to answer our share and position in the market.

The key to the common comparative application of Aamp B test,A/B test is to ensure that there is only one single variable in the two groups and other conditions are consistent. For example, to test the revision effect of the home page, you need to ensure that the source channel is the same, the user quality is the same, and the online time is the same, so that the tested data is meaningful.

III. Quadrant analysis

According to the difference of data, each comparison subject is divided into four quadrants.

Generally speaking, registered users of P2P products are drained by third-party channels. if they can be divided into four quadrants according to the quality and quantity of traffic sources, and then select a fixed time point to compare the performance-to-price ratio of each channel, the quality can be measured by the dimension of the total amount retained. For high-quality and high-quantity channels to continue to maintain, for high-quality and low-quantity channels to expand the introduction of quantity, low-quality and low-quantity pass, low-quality and high-quantity to try to put in the strategy and requirements, such a quadrant analysis allows us to get a very intuitive and fast result in comparative analysis.

IV. Cross analysis

Comparative analysis includes both horizontal and vertical comparisons. If you want to compare horizontally and longitudinally, there is a cross analysis method. Cross-analysis is a cross-display of data from multiple dimensions and a multi-angle combined analysis.

The main function of cross-analysis is to subdivide the data from multiple dimensions to find the most relevant dimensions to explore the causes of data changes.

Common dimensions are:

Time-sharing: whether the data has changed in different time periods.

Sub-channel: whether the data from different traffic sources have changed.

Sub-users: whether there is any difference between newly registered users and old users, and whether there is any difference between high-level users and low-level users.

Sub-region: whether there are any changes in data from different regions.

Cross analysis is a process from coarse to fine, which can also be called subdivision analysis.

Trends, comparisons, quadrants, and crossover contain the most basic parts of data analysis. Whether it is data verification or data analysis, finding trends, making comparisons, drawing quadrants, and subdividing data can play its due role. Hope that through the above sharing, can help data analysts to do a better job of data analysis.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.