Відео

I replaced the base64 images with external urls but those images would not load :(

Not possible. At least for now. Table, Matrix, Card (new) visuals block mouse events in SVG images... For clickable maps, use IconMap or Deneb visuals.

Thanks. Regarding the sound: I'm sorry if it's not very good, I probably shouldn't record and edit videos in the middle of the night.

@power-of-bi What could help is to create the audio stream separately. Typically a smartphone with a recording app put in front on your desk pointing to your mouth would give sufficient results. In video post production, just put this in as a separate audio layer. Works typically totally reliable and you do not need any noise cancelling / auto-ducking features in your video editing software which can also give these odd results. Thanks for the otherwise great content!! 🙏🙏

@@businessinsights_AlexRobe Thank you! The audio is from a better microphone than any smartphone has, but I missed an issue with mic settings... I believe I've already fixed the problem. Publishing a new video today...

Andrzej, you showed how to work with two datasets that originate from the same source. I wonder is it possible to work with several dataset that are not originated from the same root data source? If yes it would be great if you covered how to in your future videos.

Thanks. Really great question. Well, there is no a single way to create a visualization. It's perfectly OK to use the same data source for the rectangles and for the text labels. And in many cases there will be no meaningful difference. In this example you can replace {"data": "rect-ac"} with {"data": "table"} and then remove 'datum.' from the "field" properties within "encode", because as you have mentioned we don't use fields generated by Vega for the "rect-ac" table (such as xc, y, width, height), we still use the original "table" fields stored in the datum field of the 'rect-ac' table. Or you can keep {"data": "rect-ac"}, remove the "scale" and "band" properties and use "field: "x" instead of "field": "datum.month", "field": "y" instead of "field": "datum.AC" and "field": "datum.AC" for the "text". In more advanced visualizations using a table like "rect-ac" (a table created by adding "name" property to a mark) will be useful to avoid repeating some heavy calculations multiple times (assuming there was some tricky formula to calculate "x", "y", "width", "height", "fill" or any other property of one mark (such as "rect-ac") and you need the same for the other mark (such as "text" for the text label) . In this case it doesn't really matter. Whatever piece of code was easier to copy paste from the previously created marks and modify. Thanks again. I'll talk about this in the future videos.

Thank you very much for your clarification, Andrzej! These step-by-step videos with detailed explanations are very helpful.

This will be a subject of the next video. Thanks.

ua-cam.com/video/oun2V09Dbt8/v-deo.html

Thanks. Import the .csv data file into Power BI and load into Deneb "dataset" deneb-viz.github.io/dataset. I'll make a video.

ua-cam.com/video/oun2V09Dbt8/v-deo.html

Thank you for the comment

Thank you for the comment. Such comments encourage me to make more videos.

Vega visualizations (including interactive and animated ones) can be created using Vega JSON only. No JavaScript required. Basic JS may be required to embed Vega visualizations into a web page with some additional functionality, for example, Vega examples on my website are being generated using Vega JSON code stored in my Github repository. That required a few rows of JS code.

@@power-of-bi Great. Thank You. Hoping to learn a lot from you.

Thanks. I'll do that soon. I'll start from a simple example (basic column chart). Then, in the following videos will go through other more and more advanced examples.

ua-cam.com/video/HvIxqEC2_TY/v-deo.html

For some reason most people prefer Vega-Lite (less code?). From my point of view (but maybe it's skewed by the fact that I already know Vega) Vega is more descriptive and straightforward. I mean, look at the Vega mark. It's clearly describes what you see on the screen ("rect"), what data table it is based on ("table"), where it starts and where it ends (x, y, y2 and width properties), how values are being projected onto the plot (using xscale and yscale) and what color it is filled with ("steelblue"). Vega: "marks": [ { "type": "rect", "from": {"data":"table"}, "encode": { "enter": { "x": {"scale": "xscale", "field": "category"}, "width": {"scale": "xscale", "band": 1}, "y": {"scale": "yscale", "field": "amount"}, "y2": {"scale": "yscale", "value": 0}, "fill": {"value": "steelblue"} } } ] And Vega-Lite? "mark": "bar", "encoding": { "x": {"field": "a", "type": "nominal", "axis": {"labelAngle": 0}}, "y": {"field": "b", "type": "quantitative"} } What? Everything is based on default setting applied behind the scene.

@@power-of-bi Very good example to appreciate the difference. Thank you.

An HTML table formatted and animated using CSS, then embedded into an SVG namespace using the foreignObject element. This was implemented in the New Card visual.

Share a video how to prepare this pls

For me the best way to learn Vega was by analyzing the code of the official examples vega.github.io/vega/examples/ and adjusting the code to see how my changes affect the visualization and by reading the documentation vega.github.io/vega/docs/

Use absolute values (not variance) to calculate scale ratio and then apply this scale to the absolute variance.

I'm sorry that it takes for me more time than you would expect, but I'm actually have just started a series of video about Deneb and Vega. In the first video ua-cam.com/video/yIcBpbui7U8/v-deo.html you'll find some information about what resources you can use to learn, what people you can contact to ask for help. In the next videos I'll talk more about real examples starting from the simplest ones (so it's going to be a tutorial).

I'm glad the video helped you! Thanks for the comment.

Yes you can. See github.com/avatorl/Deneb-Vega/tree/main/ibcs-three-tier-chart

I've built an income statement with SVGs in a Matrix visual, but for now just a basic one, not a large ragged hierarchy with expandable rows. Regarding the hierarchies see www.daxpatterns.com/parent-child-hierarchies/.

You're welcome!

Thanks. That's a great question. It's possible to draw multiple bars in a cell. The question is what are you going to have in the columns. It makes sense to have a date per column, therefore what you need is a either blank cell or a cell with 100% column width bars. Although, there will be an issue, you'll have non-removable margins around the SVG images (empty space between the columns). If ability to select a single date is no important you can have put for example weeks into the columns and wider (up to 512 pixel) SVG images to draw bars with variable length (to cover start-end date period).

@@power-of-bi Thanks for the fast response, Andrzej! I really appreciate it! It seems like its possible to prepare such Gantt chart, but there are some workarounds to be considered. I am still a newbie using PBI and will investigate more according to your feedback. Greetings from Poland! ;)

Thanks for you comment. Answers to the questions: 1) DAX measure can't know anything about the matrix size (individual image width and height), you'll need to change options on the Format pane and then to edit corresponding variables in the measure manually 2) Columns width is based on Font Size. So for smaller chart you'll also want to edit font size variable. 3) Flipping between horizontal and vertical definitely requires different measures (see another video for the vertical chart). I see no problem here, they have different use cases, one for time, one for structural analysis. 4) Data labels are for AC (and FC) column, not for PL columns. I believe we shouldn't place AC data label right above PL (outlined) column. But it's OK to just hide labels in certain cases to avoid overlapping.

@@power-of-bi Add your Variance one level back. It will look a bit better: VAR _SVG = _SVG_URLPrefix & _SVG_Open & _SVG_Variance & _SVG_PL & _SVG_ACFC & _SVG_PY & _SVG_DataLabel & _SVG_PeriodLabel & _SVG_TextStyle & SVG_HatchedPattern & _SVG_Close Also for extra clarity consider setting your FC triangle black VAR _triangleFill = IF ( _MainScenario = "FC", _ColorBlack, _ColorGrey)

​@@apbreports Thanks. 1) I understand what it will change. Makes sense. Great tip. I'll try. 2) I'm sorry, I see your point (higher contrast), but this proposal contradicts IBCS recommendations (and visual perception principles). It will ruin the entire scenarios unification. Things that are the same should look the same. Triangles are grey for a purpose. Grey fill color = previous year data. IBCS, first of all, is not about 'looks better', but about 'works better'. Comparison with the triangles is secondary task anyway, they are not critically important.

Strange. I was 100% sure I've already answered here... I'll try again. Thanks for the comment. 1) I understand what you're talking about, makes sense, I'll give it a try 2) I see your desire to increase visual contrast. But that will contradict visual perception principles and IBCS recommendations, ruin scenario unification and make the chart unclear. Thing that are the same (previous year data, grey triangles) should look the same. And grey color is not random. It has a specific meaning here and shows older data (previous year) than black AC. Every color (and hatching) is meaningful.

@@power-of-bi I understand about the triangle contrast. Best to keep to the standard of course:)

Thanks. First of all, it's just a demo so far, not designed and not tested to handle all possible situations automatically. But, what you're asking about is already handled by IBCS. And it's very logical and straightforward solution - we don't really need to see more than 3 digits on data labels on a chart. 500 (k) and 10 000 (k) can be replaced with 0.5 (m) and 10.0 (m). And category with (period or month in this case) and column width (which is 2/3 of category width) is based on font size to get perfect composition of the columns and labels. a scroll bar in a chart? never!