How to Conduct Data Analysis of Subjectivity 70

Data analysis of subjectivity is statistical inference, which divides the views of the person into categories such as:

One, two, three, and so on are understandable, but counting the number of supporters of an idea is only the first step in statistical inference, and just based on the amount of these raw data (there are more people who support point one, those who support point two are second, and those who support point three are relatively small). Question) is only a general statistical description of the problem and cannot be used as a basis for statistical inference, i.e., conclusions.

The analysis of subjective data,1 divides the views of the people into categories such as:

One, two, three, etc.; (When classifying, minority views should also be listed separately) 2, and the data should be counted separately (the number of counts);

3. According to whether the data population conforms to the normal distribution, the significance hypothesis test of the difference is carried out separately. Yes, tested with parametric hypotheses; discrepancies, tested with non-parametric hypotheses. A significant difference is only true if the difference exceeds the probability at the random level.

before we can draw conclusions. The significance level of hypothesis testing "alpha" is generally taken or.

4. When using graphics to assist the image and clearly illustrate the differences between different points of view, choose a bar chart. Bar charts are mainly used to represent discrete data, i.e., counting data. The bar chart uses the length of the bar to represent the size of the quantity, and it is easy to see the number of various quantities from the bar chart for easy comparison.

For example, 100 subjective test papers.

Each answer will have its own point of view, and there will be repetition in these 100 points. Assuming that there are a total of 3 main points, the data analysis is as follows:

Opinion Proportion--- i.e., people who support that view.

Viewpoint 1 40

Viewpoint II 30

Viewpoint III 20

Other perspectives 10

Other views are those that are confusing or unmainstream or extreme.

Based on the analysis of the data, the conclusions are:

There are more people who support the first point of view, which means that ... Issue.

Those who support point 2 are second, which means ... Issue.

There are relatively few people who support point three, which shows that ... Issue.

Other points of view are briefly reviewed.

Through data capture and data monitoring, it is integrated into a huge database - industrial economic data monitoring, ** and policy simulation platform.

Carriers cooperate with accurate data sources, establish a database according to your industry, and belong to your own exclusive model to obtain high-precision intent customers. Wide gripping range and large radiation area.

The collection of big data is nothing more than the collection of data from a software ** or software system, because the data is in the database of different software, and the collection and acquisition of data need to find those software manufacturers to do the interface, which is also the current mainstream solution, and now there are some new solutions, with the 101 software interface generator as the convenience, do not call the data from the database, collect data from the software interface, and achieve the purpose of data collection.

Hui Research (professional third-party market research service provider).

First of all, let's talk about the importance of writing a good data analysis report, which is very simple, because the output of the analysis report is the result of your entire analysis process, the qualitative conclusion of evaluating a product and an operational event, and it is likely to be the reference basis for product decision-making.

First of all, have a good framework.

Like building a house, a good analysis must have a foundation and a solid foundation, and the level is clear to make the reader clear at a glance, and the structure is clear and the priorities are clear to make it easy for others to read, so that people have the desire to read;

Second, every analysis has a conclusion, and the conclusion must be clear.

If there is no clear conclusion, then the analysis is not called analysis, and it loses its own meaning, because you are going to find or confirm a conclusion before you do analysis, so don't forget the original fruit;

Third, the analysis conclusions should not be too many and precise.

If you can, an analysis of the most important conclusion is good, many times the analysis is to find problems, if one by one analysis can find a major problem, it will achieve the goal, do not seek more for everything, rather than a bite of peaches, not a basket of rotten apricots, the concise conclusion is also easy for the reader to accept, reduce the important reader (usually a leader with a lot of affairs, not too much time to see so much) reading psychological threshold, if others see too many problems, the conclusion is too complicated, do not read on, a hundred conclusions are equal to 0;

Fourth, the analysis conclusion must be based on the closely forbidden data analysis and derivation process.

Don't have speculative conclusions, things that are too subjective will not be convincing, and if a conclusion is not even sure of yourself, don't take it out to mislead others;

Fifth, a good analysis should be highly readable.

This refers to legibility, everyone has their own reading habits and ways of thinking, you will always write according to your own thinking logic, you yourself feel very clear, that's because the whole analysis process is done by you, others may not know so much, you must know that readers often only spend less than 10 minutes to read, so you have to consider who your analysis reader is? What do they care about most? You have to put yourself in the reader's shoes to write an analytical email;

Sixth, the data analysis report should be as graphical as possible.

This is actually a supplement to the fourth point, using charts instead of a large number of stacked numbers will help people to see the problem and conclusion more vividly and intuitively, of course, there are not too many charts, too many charts will also make people confused;

Seventh, a good analysis report must be logical.

Usually it is necessary to follow: 1. Find the problem 2, summarize the cause of the problem 3, solve the problem, such a process, and a logical analysis report.

It is excel significance analysis of variance, and the difference in the mean is achieved through the standard deviation, or the difference system. The formula for calculating the standard deviation can be seen by looking at the statistical function categories.

SC-CPDA Data Analysis Public Communication Platform View my profile for details.

Weibo data analysis can be deep or shallow, and if you want to speculate on the blogger's business strategy, you need to track the data for a period of time, and the minimum time is a week. Weibo to implement the real-name system, this initiative has a great effect on the healthy development of Weibo, network culture is valued, Weibo marketing is bound to receive more and more attention, the data analysis of Weibo is conducive to us to better do Weibo, so what is the specific collection of Weibo data? What can be seen in the data?

1 Fans: From the perspective of fans, people with a large number of fans can naturally attract people's attention, if the growth is fast, what can it mean?

2 Content: Judging from the blogger's Weibo content, what type of Weibo is it? Is it purely original, or is it an activity such as voting, with a prize**?

How often do bloggers post content every day? The ** of Weibo content, is it original product information or all kinds of sharing, or is it from the PP content library?

4 Attention: From the perspective of attention, what people and industries do bloggers follow, and whether they are in the same industry? Among the people who are concerned, are there many people who add V certification? What does a lot of words say?

After the collection of the above data, it is not difficult to see that the main strategy of bloggers is to increase the number of microblogs to increase the influence of microblogs. So how to increase the number of microblogs, the most important thing is the content of microblogging, judging from the collected data, the content of the activity is often very high, generally there are hundreds. For example:

**Entry Weibo and @3 friends, you will have a chance to win prizes. What's more, the blogger will put a certain Weibo at the top, then the number of this Weibo will naturally increase.

What does a Weibo private message do? Most corporate microblogs have a private message function. The author thinks it would be better to have a private message function.

Private messages are very good for the interaction between bloggers and netizens, and the new version of Weibo's private messages have a section similar to the chat window, which is very convenient for those who have used it. It doesn't have to be @who, @之后的语句是所有人都可见的 to communicate with netizens, and private messages are private. As a blogger with hundreds of thousands of followers, when you have a conversation with a very ordinary fan, the fans will be very happy to interact.

This increases the stickiness and loyalty of fans.

The development of Weibo is favored by many people, especially after the real-name system, analyzing the opponent's Weibo or the top Weibo will help us gain a firm foothold in the Weibo trend.

sc-cpda see my profile for details.

5. Help enterprises to carry out services.

Through big data computing, social information data, customer interaction data, etc., can help enterprises to carry out the horizontal design and fragmentation of brand information. Economist Richard H

Thaler once argued that "a small change in an individual's perspective can turn into a major change in the pattern of group behavior for all." "In the context of this major change, for the tiny flow of information, enterprises are.

It has to be taken seriously, and customer service needs to be like air in the minutiae in order to cope with this situation. Enterprises can use the massive data disclosed in social networking to cross-verify technology and score through big data information.

Do you understand these five roles of big data? How to make good use of big data can start with its five functions.

Big data analysis and processing is carried out through Internet information cleaning, extraction, deduplication, classification, summarization, clustering, association, indexing, and storage.

An independent analysis engine system, in which the configuration management platform module is a BS structure, and the engine tool module is a CS structure graphical user interface, which adopts multi-machine distributed and single-machine multi-instance deployment. The engine tool module is divided into four sub-engines, which run according to the data flow robot model of the data cleaning engine, the data characterization engine, the data analysis result generation engine, and the data result rendering engine.

The engine tool module studies and judges the data through text mining technologies such as automatic word segmentation, automatic clustering, automatic classification, rule classification, mixed classification, text similarity retrieval (automatic repetition), automatic abstract + subject word indexing (free word + industry subject heading), common sense proofreading, information filtering, pinyin, homophonic search, related phrase retrieval, natural language retrieval, etc., and realizes structured and unstructured data management in combination with full-text retrieval technology, and supports the mixed retrieval of structured and unstructured data.

Big data: Data collections that are difficult to obtain, store, manage, and analyze with conventional database tools.

I would like to introduce you to a big data analysis software that is easy to use.

Tempo big data analysis platform is an all-in-one big data analysis application platform for enterprise-level users. Based on the big data architecture, the platform integrates data visualization exploration, in-depth data analysis, and achievement management applications, and effectively solves the problem of data value exploration and utilization for data analysis and data value utilization personnel at all levels of the enterprise. The platform's convenient data access and preparation, integrated data mining and visual analysis, flexible and diverse achievement management and application, provide users with professional, agile and easy-to-use data analysis and application experience.

1.High-performance big data processing.

Based on the big data architecture, it supports distributed storage, distributed parallel computing, and in-memory computing to achieve massive data analysis.

2.Leading analytics algorithm engine.

Based on the original distributed algorithm engine of big data mining application, it is embedded with the world's leading l sparse iterative regression, visual clustering, sparse time series and other algorithms, as well as self-developed Chinese text algorithms.

3.Flexible and open system integration.

The flexible and open architecture supports the rapid expansion of graph and algorithm nodes, and supports seamless integration with the existing business systems of the enterprise.

Exin ABI is a set.

A platform for data collection, processing, analysis and display. It fully meets the user's data application scenarios, and provides users with a one-stop data analysis platform through rich data analysis methods. And to a large extent, it can reduce the technical threshold of data analysis implementation, simplify complex work, and make repetitive work intelligent.

The data integration module in Yixin ABI is equivalent to a complete set of data warehouse implementation tools, in which the rich processing and conversion components realize data extraction, cleaning, conversion, loading and scheduling through drag-and-drop process design, which is used to help enterprises build data warehouses, complete data fusion, improve data quality, and serve data analysis.

There are hundreds of visualizations and graphs built into report analysis. It not only supports more than 80 kinds of statistical charts, but also includes maps and GIS maps of the world and various provinces and cities in China, and thousands of visualizations can be derived through design and collocation. At the same time, ABI also supports dynamic and cool cool screen analysis, unique 3D panoramic perspective, and freely and quickly produces various interactive conventional screen and large screen reports, turning creativity into reality.

At present, the leader in data analysis in China is FineBI, and multi-dimensional OLAP analysis is the concentrated embodiment of the analysis function of BI tools, and its application characteristics are mainly reflected in two aspects: one is the real-time query effect (online), which requires the calculation speed of background data and the display speed of the front-end browser to be very fast; The second is multi-dimensional custom analysis, which requires that the multi-dimensional database of BI tools should have greater flexibility and can combine arbitrary indicators and dimensions according to user requirements. Only the interactive analysis process that satisfies these two characteristics at the same time is multi-dimensional OLAP analysis, which can ensure that users can see the statistical results corresponding to their analysis needs in real time, and meet the new analysis needs generated immediately according to the results of the previous step by switching dimensions and changing conditions.