Showing posts with label big data. Show all posts
Showing posts with label big data. Show all posts

Tuesday, January 09, 2018

Trouble Digger

It is always fun on discovering bugs or flaw on software, not because it is showing how good I am but rather it makes software better, giving users better experiences.

I am not software developer nor I am good in logic. But the challenge of works is always about :"how do you sure the software is running as expected?" Software may have been tested based on hundreds of test cases, but there is always some cases will be missed out.

Most of the cases, only products deployed to market,  only then will able to tell the real result. And most of the time, only faults reported by users only then can check further.

I found it fun but headache too when need to check the integrity of data. Fun because it challenges my understanding of the system and algorithm. Headache because dealing with hundreds or millions of records, which is dramatically slow if check thru manually.

Luckily with the help of data analysis tools like python,  pandas,  excel, et cetera,  have make work much more easier, and fun. Basically just need to develop algorithms,  load the data in,  test and check. Of cause developing the algorithm process takes time too, most of the time have too test and re-test to make sure it covers all possible scenario.

No one like software bugs,  nor do I. Users experiences and trust is always the priority,  that is why software must be tested thoroughly to minimise the bugs, enhance user experience. When there is bugs, must be fixed as well without delay.

Do you know most users do not feedback their experiences on using products? They may face bad user experiences, either because of the UI, the process, or facing bugs, they just endure with it. But, these users are tend to share their experiences with friends, which giving bad impression of the products.

Thursday, July 03, 2014

数据

数据,可以是很骗人的东西。

多年的工作经验,面对很多很多的数据,头痛的,是要如何分析海量的数据,化为参考值。这之间,整理出来的参考值,可以是很主观的,也可以是很客观的,一切在于想呈现的是什么,最终想得到的反应是什么。

数据越大,收集的层面越广,得出的参考值,当然更准确。但很多市面上的参考值,其实都很待考证的。比如,有些报告得出的结论,都只不过参考了约千人意见而已。但事实上,千人的意见,也只不过是一个社区里的百分之一人群而已。这样得出来的数据,它的准确性, 代表性,又是多少。


面对海量的数据,如何客观的分析,如何避免在无知的情况下,陷入参考值的误区,是令人头痛的。

至少,我是头痛的。