Original subtitles

Instructor: Let's take a few minutes

and talk about an extremely important topic,

data visualization best practices.

Now, data viz is equal parts art and science.

And before you start just dumping data

onto a canvas or choosing whatever chart looks pretty

or happens to fill the page,

you need to step back and ask yourself

these three key questions.

Number one, what type of data am I working with?

Is it geospatial?

Is it time series?

Are there hierarchies?

Is it financial?

And so on.

Number two, what exactly am I trying to communicate?

Right?

Is it a comparison across categories?

Am I trying to show a composition,

a relationship, a distribution?

And third, who is my audience?

Right?

Who's the end user and what exactly do they need?

Am I presenting to or designing for a fellow analyst,

for a manager, for an executive,

or even for the general public?

That end user will really dictate how my visuals

are designed and developed.

So let's go ahead and unpack each

of these questions in a bit more depth

starting with question one,

what type of data are you working with?

Now, data comes in all shapes and sizes.

It can fall in all sorts of categories.

And some of the things you're looking for

are things like, do I have time series data?

Right?

Is there a date field that lets me show trends

or patterns over time?

Do I have geospatial fields that let me draw comparisons

between geographic regions

or locations using things like maps?

Are there interesting categorical fields

that I can use for filtering

or segmenting the data in my reports?

Again, do I have hierarchies that I can drill up

or down into as part of my analysis?

And then there's some less common categories as well.

You might have financial specific data.

You might have text data

that's a little bit less visual or numeric.

You may have funnel stages represented in your data set

or even things like survey responses.

And again, there are many, many more examples

of the types of data that you might encounter.

But the bottom line here is that the type

of data that you're working with will often determine

which type of visual will best represent it.

For example, using maps to represent geospatial data,

using line charts for time series data,

bar or column charts for categorical comparisons,

tree maps for hierarchical data, and so on and so forth.

Question two is all about what you're trying to communicate.

So let's break this down

into four different categories here.

Could be a comparison, a composition,

a distribution, or a relationship.

Now a comparison is when you're trying to compare values

either over time or across different categories.

And the common visuals that you'll use

to communicate comparisons are things

like basic column and bar charts, clustered columns,

data tables or heat maps, if you're using time series data,

line charts or area charts.

And then sometimes more specialized visuals

like radar charts can be helpful here as well.

Composition is all about breaking down

the component parts of a whole.

This is where you'll typically use visuals

like stacked bar or column charts, pies or donut charts,

stacked areas to show both composition

and trending over time,

or possibly some more specialized visuals like waterfalls,

funnels, tree maps, or sunbursts.

Distribution is about showing the frequency

of values within a series.

And histograms are really far and away the most common

and popular type of visual to show distributions.

If you've ever seen a bell curve or normal distribution,

that's a histogram at work.

You might use things like density plots

or box and whisker charts here as well.

And last but not least, we have relationships

which are about showing correlation

between multiple variables.

Scatter plots and bubble charts

are far and away the most common visuals in this category,

could also potentially use data tables,

heat maps, or a correlation matrix as well.

So this can be a handy guide to help point you

towards the right visuals to choose

based on what you're trying to communicate.

And I know this is a quick review.

I know this can feel a little bit overwhelming

talking about all of these different chart types.

How could you possibly know which one

to choose for any given situation?

So the big takeaway here,

the bottom line is to keep it simple.

There are hundreds of charts,

if not thousands of charts to choose from.

But at the end of the day,

your basic tried and true options like bar charts,

column charts, line charts, histograms, scatter plots,

those are gonna be a great fit in 90% of use cases.

And the reason they're tried and true

is because they often do the best job

telling the simplest and clearest story

which ultimately is the goal of data visualization.

And that brings us to question three.

Who is the end user and what do they need?

So let's simplify things a bit and imagine

that there are three potential audiences.

We've got the analyst, the manager, and the executive.

Obviously, there are different variations

of end users and audiences that exist out there

but this will help us start to understand

how to tailor an analysis or visualization

based on who's consuming it.

So when you're designing for someone at the analyst level,

typically these are people who like to see details,

they want to understand what's happening

at a granular level, they're analytically minded,

so they might want to see things

like tables or combo charts, a bit more complex,

maybe a little bit more data heavy.

And again, they'll want access

to some granular detail to support root cause analysis.

On the other hand, a manager level audience

might want more summarized data with a focus on clear

and actionable insights to help them operate the business.

So in that case, it typically makes sense

to skew towards more common

or basic charts and graphs, some detail,

but really only when it supports

a specific insight or recommendation.

And then finally, at the top of the food chain,

you've got the executive audience.

These are people who are super busy,

typically they just want high-level,

crystal clear KPIs that they can use

to track business health

and top-line performance at a glance.

So this is where visuals like KPI cards

or very simple charts often make the most sense,

and you wanna keep the detail

to a minimum unless it adds critical context to those KPIs.

So again, takeaway here is that how you visualize

and present your data is largely a function

of who will be consuming it, right?

So that fellow analyst might want the granular details,

managers and executives might prefer top-line KPIs.

And again, that focus on clear data-driven insights.

So there you have it.

That's our crash course

in the three key questions for data visualization.

Next up, we're gonna talk

about our step-by-step dashboard design framework

and then we're actually gonna put pencil to paper

and start sketching out the layout

for our Adventure Works report.

Stay tuned.