In the era of universal BI tools, can chat records be used for data analysis?

Recently, there is a problem on Zhihu:

After reading some of Gao Zan's answers, my biggest feeling is: "Thailand, Singapore, Indonesia" before marriage, "after marriage" toys, kindergarten, all are babies. As a school-age young man, he suddenly became curious about his life after marriage, so I compiled the WeChat chat records between me and her.

Through the analysis of "big data", let's see what has changed in the "married" life?

--data preparation--

Since living together in January 2017, life has been the same as being married. Therefore, set the data of "September 15-17 January" as "before marriage", and set the data of "January-18 September" as "after marriage", and analyze the total Three years of data, a total of 135,687 items, let’s take a look at the difference in life before and after "marriage".

--Analyze content--

1. Daily trend of the number of conversations

In the era of universal BI tools, can chat records be used for data analysis?

After "marriage", there was a Bitcoin-like drop in the number of conversations. The average number of conversations after marriage was compared to the peak period before marriage. When you see this curve trend, you can't help but not draw a pressure line on it. , The so-called perfect trend is probably the same. Every time it rushes higher, it does not break through the pressure level. Recently, there is a stabilizing trend, and it is expected to continue to oscillate in a narrow range. Just as Bitcoin now accounts for the majority of over-the-counter transactions, most of the chats in post-marriage life have migrated offline. Of course, the bull market of Bitcoin will come again, but the number of chats is expected to fluctuate within a narrow range. The bears still alternate, but they get married only once!

2. The number of conversations within the week

In the era of universal BI tools, can chat records be used for data analysis?

The number of conversations "before marriage" follows the "Lost Theorem": the number of conversations is proportional to the degree of miss, and the degree of miss is proportional to the time since the last meeting. No see in one day, like three autumns. The number of conversations "after marriage" conforms to the principle of conservation of volume: the total number of conversations is conserved at the same level of thoughts. See you every day after marriage, and my thoughts turn to be dull. After I finish talking offline, I won't talk online.

3. Time-sharing trend of the number of conversations

In the era of universal BI tools, can chat records be used for data analysis?

"Before marriage", life is very regular: busy at work during the day, chatting happily before going to bed. "After marriage", life is also regular: busy at work during the day, and at night, either overtime or meeting.

4. Proportion of dialogue content types

In the era of universal BI tools, can chat records be used for data analysis?

Things that could be solved by selling cute (animated emoticons) before, now have to use money (red envelopes).

5. Chat content analysis

In the era of universal BI tools, can chat records be used for data analysis?

"Before marriage", a collection of WeChat intimate expressions: kiss, hug, sun, good night, kiss

"After marriage", a hodgepodge of trivial matters in work and life: overtime, noon, going home, cafeteria, tidying up

--Analysis conclusion--

Before marriage, there are many and sweet talks, emotional ups and downs, and they all talk about the taste of love;

After marriage, the words are few and true, although they are plain, they are all about the beauty of life!

The summary is:

More romantic before marriage, colorful after marriage

(For this reason, in all the previous pictures, "before marriage" uses romantic purple, and "after marriage" uses colorful colors)

It can be seen that data analysis is not only in boring work, but also can go deep into life and run through. Data analysis is easy to get started, but deep learning is difficult. Software choice is the key, Smartbi is the first choice. You can try to use Sematic software products for data analysis. Based on Excel, it is easy to learn and everyone is a data analyst.

This article is reproduced in the public account: Two Officials and Two Listening

Author: Liang, Internet data analyst.

If there is any infringement, please contact us to delete it immediately, thank you.

Guess you like

Origin blog.51cto.com/15081336/2634144