Data analysis for game operations

I've been looking at how to start a data analysis article these days, everyone writes well, to tell the truth, for how to launch the game Operation Data Analysis Guide really very few, as each company's core secret is not to get the table to share with you, a period of time I read a lot of materials, when the display can not meet the demand, We'll have to dig on our own. Here is my summary of the data analysis method after combining some articles.

———————————— Split Line —————————————————

The data analysis work can be from macroscopic data and microscopic data (subdivide data) to talk about, this way also I like, as Xiao Qiang said, macro data is the overall trend of the forecast, and the sensitivity of the anomaly data grasp. The source of microscopic data analysis is the demand from the macroscopic data, which is a kind of mutual dependence. Of course, if not professional to do da work, perhaps in this way there is no problem, because after all, work time and energy does not allow more research work. From my rookie da, there is still a lot of work to be done, and the form taken is another one, but the meaning is the same as before.

The following figure, probably every industry's data analysis system is this pattern:

The regular data of the online game to grasp and detect more is for the popularity (total login, peak, APA, registration, loss, online hours), consumption (ARPU, recharge, consumption, permeability).

Thematic data mining is currently used in the field of data analysis of online games is relatively small, even if this research is the core technology of the company, this part of the research is on the whole game player's game behavior, purchase behavior, emotional behavior, game psychology, game pressure, game life, game experience, game interaction, IB Purchase Association preferences, Economic system operation analysis and so on in-depth research, not to solve a problem, but a regular thematic research based on mass data analysis, only a deep understanding of the needs of users to make and operate well in line with the players taste products.

User Research in fact, in the online game data analysis work on the edge of the position, many players do not know what they want, so to some extent, we do this kind of research work will often get the wrong player signal, so rarely use research methods to analyze the player.

In this way sum up the following figure:

The depth finder here is a long-term and fixed search for various characteristics of the user.

So for us, there are two pieces of work, regular data analysis, thematic data mining research. Conventional data analysis, in addition to macro-grasp of data trends and differences, but also in the micro-dynamic data indicators to subdivide, from a microscopic point of view to find out the problem. And the topic of data analysis is our initiative to ask some questions, and then to find data and research, not to solve the problem. This does not seem to be the most direct solution to the problem, however, the interpretation of these data, we can grasp

What the player wants (what);

Why (why);

Where to get (where);

When do we do it?

Which players are targeting which operational strategies (WHO);

How much we should give;

in what form;

Through the 5W2H method, combined with analytical means to solve these problems. The following is a summary of the network based on the data analysis of some points of attention and methods.

The idea of conventional data analysis--from the perspective of income

But when we are faced with a decline in earnings, we need to locate the problem and solve it from the perspective of profitability.

The idea of conventional data analysis--from the perspective of popularity

Through the above data interpretation and the segmentation of these macroscopic data, we can complete the analysis of some abnormal data and urgent needs.

While doing well in this work, we also need to do a good job of data analysis, provide operators more operational decisions.

For the special research of game data mining, the following points are summarized as follows:

In the topic of data mining and analysis patterns, there are several forms:

• User Lifecycle Model

• Loss factor function and model calculation

• Network Media effect analysis

• Game activities and system risk assessment

• Game economic system Early warning assessment

For the thematic type of data mining, is still in a slow process of research, this piece of really

The

is more difficult, unlike the traditional retail, financial, and telecommunications industries. The net swims has the uniqueness in the concrete analysis process, needs to combine the characteristic, the reasonable application theory and the technology solves the question.