All language subtitles for 15. Power BI Demo Adding Trend Lines & Forecasts

So in the last lecture we've built out this line chart and this area chart or product detail page but

I feel like our line chart is kind of sad.

It's shown profit by week which is helpful but I think we can pack some more information in here.

And one way to do that is to use the tool tips like we've practiced already.

So let's go ahead and grab some important measures like revenue and orders or tool tips that way all

three of those metrics are now visible at the weekly level without having to chart three separate series

on the same graph which would certainly be feasible but not the best visuals since those metrics are

on very very different scales.

Instead what I want to do is explore a pain that we haven't touched yet which is this third option for

analytics.

Now this pain will only be accessible for certain visuals and for many these options will be quite limited.

But the line chart is one type of visualization that has very rich very robust analytics options.

So we're in a play with a couple of these here starting with the trend line.

Now all we have to do is add a trend line and it essentially fits a line to our series over time to

give you an overall kind of smooth out trend of where things are headed.

So if I'm Adventure Works This is a great trend because I'm seeing profit on a nice upward slope over

the last year and a half or so.

Now like everything else the look and feel is completely customizable.

This case that black line is pretty overpowering.

It's going to pull that down to about 75 percent transparency.

That's a little bit more subtle which I like you can change the style to solid or dotted or dasht.

I think I actually prefer that dasht here.

And then these default options will do just fine.

These come into play if you're dealing with multiple series.

So that's our trend line.

Easy enough.

We have very similar options for a constant line a min max an average median or percentile.

All of these essentially add lines horizontal lines to our chart and they kind of speak for themselves.

Constant allows you to define a value like 100000 for instance.

And I told him you go a hundred thousand and plot that horizontal line on your chart.

That's nice if you have like a fixed goal or something like that.

You can also plot the minimum line which finds the smallest or lowest value in the window of time and

applies a horizontal line based on that minimum.

Same story holds for Max up at the top.

Kind of gives you a good sense of the range of values that you're looking at over this period.

And then average median percentile.

They all work exactly the same way.

I don't want really any of those we could add them but I feel like it's just going to make things a

little bit too busy in this case especially because I want to play with one more tool which is the forecast

down here at the bottom and this is how we're going to address that question for Adventure Works about

how is my profit expected or anticipated to look for the next X weeks or periods or months.

So Part II makes it very user friendly to create forecast views like that.

Now we're going to do is add a forecast by default created a forecast of 10 points and points are basically

date fields.

And what point is the default unit of date that you're looking at.

So in this case it's 10 weeks out and if you want to get totally wacky you could do something like you

know show me a forecast for 10 years.

And it's it's very ridiculous because you very quickly lose confidence in what your profits are going

to look like 10 years out.

So I would not recommend that and I don't think our clients would appreciate that much either.

So let's stick with weeks and let's actually narrow it down a bit to a six week forecast.

Now you do have an option to ignore previous points like if this drop if we know this was an outlier

or an anomaly we could ignore that one point.

I kind of feel like that's sweeping things under the rug so let's keep that at zero.

And we'll keep the true forecast here.

Now confidence interval this is an interesting one.

And this basically defines at what level of confidence do we think the true values are going to fall

into the shaded range.

So at 95 percent that's pretty darn confident.

Which is why you see a pretty broad or pretty wide range here.

If we drop this down to 75 and apply see that it shrinks down quite a bit and that's because we're trying

to hit a tighter target.

So naturally we're less confident that the numbers will actually fall into this narrower range.

So best practice is generally to use 90 or 95.

Let's go with 90 for our purposes there we go.

And you can also apply a seasonality factor like if you know that mid June or July profits tend to dip

over the summer.

Then you can kind of manually tweak that forecast window yourself.

So there you go.

That's our forecast.

I'm happy with that.

Feel free to adjust the settings kind of customize or personalize it as you see fit but for our purposes

I think that does a nice job it adds some good insight about the overall trend of profit and what we

might expect to see for future profits over the course of the next six weeks.