Data Analytics

with all the hype with data analyst, does anyone know how much is the pay for an entry level position? a number of companies in different industry are hiring them, but I wonder how important is the domain knowledge? It seems like the various job advertisements despite in different industry all looks the same and emphasise on the technical skills set rather than the domain knowledge. I wonder how portable are these skills gained. could anyone in the industry shed some light?

my view:

analytics is a very wide spectrum.

u have data modellers all the way down to data scientists

all are trained in different aspects. u cant expect the data scientists to know the IT stuff right.

but most jobs now are like trying to hire a 1 size fits all solution

i still think this big data is just another hype.

analytics has been ard for so bloody long

Data Analytics isn't new. It's just that because of the ultra-connected world we live in today, there are much more data now that we can play with. The underlying principle behind data analytics is pure statistics. However, people tend to forget that correlation is not equal to casuality. Case in point is the recent US elections.

Why would data analytics be so easy to enter as mentioned in this forum?
A data analyst would have to have some decent programming skills, statistics knowledge, computer science knowledge.. it's not easy to find someone with such skills except a computer science or stats students?? And roles like data scientists requires knowledge of advanced techniques.. or is it business analyst that u guys are mentioning? It seems like a lot of ppl are lumping data analysts and business analysts together in Singapore. Anyone with experience care to provide some insights?

Originally Posted by Unregistered

Why would data analytics be so easy to enter as mentioned in this forum?
A data analyst would have to have some decent programming skills, statistics knowledge, computer science knowledge.. it's not easy to find someone with such skills except a computer science or stats students?? And roles like data scientists requires knowledge of advanced techniques.. or is it business analyst that u guys are mentioning? It seems like a lot of ppl are lumping data analysts and business analysts together in Singapore. Anyone with experience care to provide some insights?

Analytics is an umbrella term. There are extremes from business analytics all the way to data scientist. That is not including other things like data architect and visualization. There isn't really on consensus on what exactly is data science, so I am not getting into that debate.

However analytics itself is simple, you need to understand basic statistical principle. You need to be logical. You need to be able to pick up new programming language quickly. And you need to be able to understand the domain. This is what is required to get hired.

In IT space, skills is ever changing. A good IT manager will hire you for your learning agility and potential, not the skill set you have right now.

A huge amount of people I know from this industry did not start from either a CS or major stats background. I've see people with engineering, business or even science degree getting into this industry.

Originally Posted by Unregistered

Data Analytics isn't new. It's just that because of the ultra-connected world we live in today, there are much more data now that we can play with. The underlying principle behind data analytics is pure statistics. However, people tend to forget that correlation is not equal to casuality. Case in point is the recent US elections.

Pure statistics is exactly that, statistics. The point of analytics is to make sense of the data and stats, not just reporting the number. Of course correlation does not imply causation, but that is where the art of data science come in with domain knowledge.

US election suffers more from unclean data than problem with the analysis itself. Trump is a controversial character and out right support of him is rather frowned about especially in the educated circles. People are not going to answer the polls honestly for fear of being judged.

For more interesting use of data science, you can refer to uber, grab, Amazon, netflix or even more recently how govtech found the rogue train using data science

Originally Posted by Unregistered

with all the hype with data analyst, does anyone know how much is the pay for an entry level position? a number of companies in different industry are hiring them, but I wonder how important is the domain knowledge? It seems like the various job advertisements despite in different industry all looks the same and emphasise on the technical skills set rather than the domain knowledge. I wonder how portable are these skills gained. could anyone in the industry shed some light?

Ranges from low 3 to high 3.

Domain knowledge is important. How are you suppose to provide recommendation and insight if you do not know the domain you are analyzing. However entry level jobs are not going to require you to have domain knowledge for the simple fact that you are a fresh graduate. A more mid level role would require domain knowledge.

Skills like stats, machine learning, algo programming are highly portable. Program specific skills are less so, due to the wide varieties that companies tend to use.

Just don't buy too much into the hype. If you are chasing after money, this is not it. But if you are looking for autonomy and job satisfaction, you are welcomed to try it out

Originally Posted by Unregistered

... Of course correlation does not imply causation, but that is where the art of data science come in with domain knowledge.

Woah talk talk drop new term then never give example. Got take any NUS stat class to explain lei?

Correlation of X and Y of course not causation what ... even IT manager will say that. But I don't think they themselves can explain causation. Lol.

Then correlation of X lag 1 and Y implies causation? Seems correct right? But can argue Y causes X lag 1 also, i.e. EXPECTATION of Y affects the current value of X.

But is there a different between correlation of X lag 1 and Y and regression of Y against X lag 1? Who can raise up hand and answer?

All come here talk big about data analytics let's see who can step up and really explain causation. Lai lai. If can explain then you can apply for data analytics job.

Originally Posted by Unregistered

Woah talk talk drop new term then never give example. Got take any NUS stat class to explain lei?

Correlation of X and Y of course not causation what ... even IT manager will say that. But I don't think they themselves can explain causation. Lol.

Then correlation of X lag 1 and Y implies causation? Seems correct right? But can argue Y causes X lag 1 also, i.e. EXPECTATION of Y affects the current value of X.

But is there a different between correlation of X lag 1 and Y and regression of Y against X lag 1? Who can raise up hand and answer?

All come here talk big about data analytics let's see who can step up and really explain causation. Lai lai. If can explain then you can apply for data analytics job.

You sound like someone fresh out of a stats 101 class trying to show off what you just learnt and I am not sure if this is some rhetoric to prove your superior statistical knowledge but I shall entertain you non the less.

All example you have gave is still statistics, which is exactly my point. using statistic itself will not be able to prove the causation and is not analytics, hence my point on domain knowledge.

In real world analytics, if you try to dive into data without a direction you would of course not be able to find anything. Analysis usually starts with a business question and a hypothesis you have that you want to test. Setting the right question and hypothesis would allow you to pull the right data to solve the right question


When we talk about correlation and causation in analytics, generally we are speaking about not missing out external factors that might be the root cause.

An example would be that you notice that as price of a certain product that you are selling decrease, volume sold decreases along with it. If you were to take this as causation, you would wrongly assume that this is a luxury product and a price drop decrease its perceived value.

However if you look further, you might notice that the economy is entering a recession, hence leading to both the decrease in your product price and a decrease in sales due to people not spending.

In fact, i would go further to say that a causal relationship can not be determined purely through statistics but through strong experimental design. In an ideal world controlled experiments would be the best way to prove causal relationship, but that is generally not possible for most industries.


If you think that hardcore statistics is the base requirement to work in majority of the analytical field, you are going to be deeply disappointed when you come out to work. Analytics uses a wide variety of tool depending on the problem you are solving, and sometimes visualization and trend identification can be just as important as any statistical test you can do. An example would be govtech's recent work on the rogue train, where they use visualization and trend identification to identify the problem before forming a hypothesis and using a basis cluster analysis to prove their hypothesis.


s://blog.data.gov.sg/how-we-caught-the-circle-line-rogue-train-with-data-79405c86ab6a#.cgb2zqkr8i