馃攳
Data Analytics - Part 1 | Fundamentals of Internal Auditing | Part 42 of 44 - YouTube
Channel: unknown
[0]
Welcome, I'm Hernan Murdoch.
[2]
In this episode, we're going to take a look at data analytics.
[12]
A common technique internal auditors use is data analytics,
[15]
and to better understand its definition,
[17]
its meaning, its uses,
[19]
we have with us Kathleen Crawford
[20]
who will help us better understand
[21]
these aspects of data analytics.
[24]
- Welcome, Kathleen.
- Hi, Hernan.
[26]
So, can we start with the definition?
[28]
What is data analytics all about?
[30]
I'd be happy to start there.
[32]
It's really, really fascinating.
[36]
Data analytics is a process
[38]
of using different computer learning techniques
[41]
to help us automatically analyze
[43]
and extract knowledge from the data.
[46]
So, if you look a little bit farther down on the graphic here,
[50]
there's a progression that we're interested in.
[53]
Data has a lot of hidden information in it,
[58]
and so what we're trying to do is to pull that information
[61]
out of the bits and bytes of the data,
[63]
and then take a look at that information
[65]
and transform it into knowledge.
[68]
So auditors will use different techniques
[70]
that uncover different patterns that might indicate
[73]
that processes are broken in some way,
[76]
not operating as expected or intended.
[79]
And we can also use it for developing some predictive patterns
[83]
in the business information.
[86]
So, instead of looking at a sample of transactions
[89]
and drawing inferences to the entire population,
[92]
now, we're looking at the whole population
[95]
and we're able to slice and dice the information
[99]
and get better information for the business.
[102]
So the better you understand the business,
[104]
the better your data mining will be.
[106]
It starts with some fundamental knowledge,
[108]
or it has to start with some fundamental knowledge
[111]
of how the area processor system works.
[114]
It also means that you have to engage in some partnership.
[117]
So if we are fortunate enough to have some data mining experts
[121]
in house in internal auditing, that's super beneficial.
[125]
But if you don't have that,
[128]
then reach out to data mining experts
[130]
as business auditors and partner with them
[133]
in order to make a more meaningful foray into this territory.
[138]
So, data analytics helps auditors in many ways
[142]
to make their results more conclusive and more convincing
[145]
because they are not only looking at a sample of the population,
[148]
but something that approximates or is the entire population,
[153]
however many transactions that will be,
[154]
it could thousands, it could be hundreds of thousands,
[157]
it could be millions of records,
[159]
so we don't need to extrapolate, we'd reviewed all of them.
[162]
So what does this process entail?
[164]
So now that we understand its power and its definition,
[168]
so what does it entail as you embark on this journey?
[171]
Well, this is a very, very systematic and disciplined process.
[178]
So we're going to talk about data analytics
[180]
from the very beginning.
[182]
And the process that I'm about to describe is called
[184]
knowledge discovery in databases, or KDD.
[188]
This is the process model that we're exploring here.
[191]
So the first thing is identifying a goal.
[194]
What is it we're trying to accomplish?
[196]
What do we want to know?
[198]
What do we want to investigate?
[201]
We might not yet be at a point where we know what we know,
[205]
but we could at least or we can't identify what we want to know,
[209]
but we can at least identify a goal.
[211]
Let's get to the bottom of this particular issue
[214]
or this particular problem, if we can.
[216]
The second step is to create or select a target data set.
[221]
So this is an opportunity to experiment and see
[225]
if we can't again, kind of bore a hole into the population,
[231]
into the set of data that will be meaningful.
[236]
The third step is data pre-processing.
[239]
Pre-processing means smoothing out what's called noisy data.
[245]
There may be missing pieces of information.
[250]
So think about this, think about some of the codes
[253]
that we expect to see for certain types of transactions
[258]
and some unusual coding could upset the truth
[263]
and the completeness of what we're looking at.
[266]
So pre-processing, the data is really, really important.
[269]
We're trying to fill in some gaps in the information
[273]
we're trying to make the information more consistent,
[277]
the data more consistent.
[278]
And so this pre-processing can be a time consuming part of this,
[284]
but it really is necessary
[286]
in order to get those really meaningful inferences
[289]
that we've talked about already.
[292]
And in some cases,
[293]
internal orders will need to merge multiple files
[295]
because they may be extracting data from multiple systems
[298]
and in a need to merge the files
[300]
or do some cleanup on the field level as well,
[304]
in terms of the structure.
[305]
For example, as it relates to dates,
[307]
something that's so much simple on the surface.
[310]
But I've done a lot of work where I had to collect data
[313]
from international locations,
[315]
and even the structure of the dates may vary
[317]
between the date month or month-date, and then the year,
[321]
and whether sometimes the year may be in the front
[324]
or the back end of the sequence of digits.
[327]
So, even those type of things may require some work
[330]
just to make sure you get everything in the same structure,
[332]
and then you can proceed with your analogy.
[334]
- Exactly.
- Very good.
[337]
So what are some of the next steps that we need to consider?
[340]
So the next thing is what it's called data transformation.
[343]
It's similar to what you were just saying
[346]
is that we need to pre-process, to fill in the gaps,
[350]
and to make sure that we understand
[352]
what each of those digits might mean,
[356]
but then we have to transform it into something that's usable,
[359]
which is often, as you say,
[361]
flipping those dates is a great example.
[364]
If we want everything to be the same from all over the world,
[367]
then we have to take some of that
[369]
and flip it to the other way that it might look in another file.
[375]
And then we mine it.
[376]
So you see that we have to go through a lot of work
[380]
before we're actually able to mine the data
[383]
and get those inferences.
[387]
We definitely need some help with interpretation and evaluation,
[391]
and this comes with experience,
[393]
or it comes with the partnership
[394]
that I mentioned on the prior graphic.
[396]
That partnership with people
[398]
who are knowledgeable about the data
[401]
and especially people who are knowledgeable
[402]
about the area processor system is what is needed here.
[407]
And then taking action
[409]
on whatever turns up in the information.
[412]
So after the interpretation and evaluation,
[416]
then taking action
[417]
into terms of making some kind of recommendation,
[420]
what you would articulate to the client
[423]
in terms of whether this area processor system
[426]
is working very well in this particular aspect,
[429]
or whether it needs to be adjusted in some way going forward.
[435]
Kathleen, this is great because this helps us better understand,
[437]
not only the definition, but also the process,
[440]
and there's a systematic way we can go about
[442]
leveraging the power of data analytics
[444]
so that we can go in a systematic way and make sure
[448]
that we get all the benefits that this encapsulates.
[450]
So thank you very much.
[451]
I really appreciate your help with this today.
[453]
It's my pleasure.
[455]
If you like this series,
[456]
subscribe to the ACI learning YouTube channel
[459]
to enjoy more audit related content.
Most Recent Videos:
You can go back to the homepage right here: Homepage