5 Quick Ways To Speed Up Your Power BI Dashboard

-




                 Power BI is a popular (sort of free) tool to create dashboards from your data. The drag and drop interface makes it quick and easy to build a simple dashboard from scratch. However, once things get more complicated and you keep adding widgets, your dashboard’s speed might get reduced to a crawling pace.

Even though there are many ways to improve your dashboard’s speed, I listed 5 quick fixes in this post that you may immediately apply to quickly improve performance without any drastic changes.

1. Use Integers Whenever Possible:

        Searching through rows of numbers is way faster than searching through strings. So whenever possible, make use of integers. For example, it would be better to use 0 / 1 values instead of “Yes/No”. Or 1/2/3 instead of “low/medium/high”. And pay attention, because a digit is not necessarily an integer. You could have 1/2/3 as values, but they can still be considered strings. So always check the data format in the query editor.

2. Remove Anything That Is Not Used:

        Don’t make your tables look like this, if there is no need to. Whether it’s tables, columns or rows, if you are not using it, it should not be in your dashboard. So during data (pre)-processing stick to what you need for your dashboard, everything else will only slow down performance.





3. Stick To The Default Power Bi Widgets:

The Power BI market place.

        The Power BI marketplace is a great resource for finding amazing widgets to present your insights. Even though sometimes you have no choice but to use them, I would advise you to stick to default widgets if you have the choice. If that fancy third-party widget looks slightly better, but slows down your dashboard, you should ask yourself the question if its worth implementing it. I’m not saying all marketplace widgets are slow, but some of the biggest speed improvements I made were by switching back to default slicers.




4. Use Tabs

         Using tabs will not only improve performance, but will probably also decongest your dashboard. If you find yourself overwhelmed by the number of slicers, charts and cards on your dashboard, it might be wise to split them into themes/subtopics etc. and place them in different tabs. By connecting your filters over different tabs you can keep your slicers, making tab A filter tab B or vice versa. The dashboard will be faster AND better looking.

5. Use Top-N Rows In Tables

        A big table can make the fastest dashboard crawl. Every time you use a slicer, the table will have to generate and display all the data in the dataset. For this reason, Microsoft added the convenient “top-n rows” feature. When activated, instead of showing the whole dataset, the table will limit itself to N rows. Again, a very simple intervention that will yield great improvement.

To Measure Is To Know

    Power BI’s new feature: The performance analyzer.


Microsoft added a new feature in it’s latest (May 2019) Power BI update: the performance analyzer. This new tool lets you analyze the speed of your dashboard and will point out the bottlenecks. It works by recording your actions when using the dashboard and timing the load times for each widget. Once you stop the recording you get a nice overview (which you can export) with a loading duration for each widget.

Comments

Post a Comment

Popular posts from this blog

4 Ways to Transform Data in Power BI Desktop

-
Image
  While making transformations in Power BI Desktop there often arrives a question which is the best method/place to do the transformations? Transformation method you choose becomes critical if you are dealing with large data sets (million+ rows) and can impact performance of your model and overall user experience. Below are some of the methods you can adopt. 1. Relational Database: If you are connected to any of relational databases like Microsoft SQL Server, Oracle DB, etc and you have the access to make transformation within the database then it is the best place to do the transformation before loading that data in Power BI environment. 2. Native Query (within Power Query): Another method which is less talked about is using native query. • Using this method you can make the necessary transformations using SQL script. • The native query fetches the transformed data directly from the database(like SQL Server, Oracle) hence avoids/minimizes your transformations in Power Query. 3. Po...
Read more

10 Pandas Functions for Faster Data Science

-
Image
  1 . read_html() Web scraping is one of the key processes that brings people to Python. Lots of people don’t know that Pandas has a web scraping function. With read HTML, all you have to do is pass in the name of the URL and you can access that data on that web page. Here is the  full documentation .                                  pd.read_html("URL") 2. corr() This function will return a correlation matrix for pairs of numeric columns.                                  df.corr() 3. drop_duplicates() Getting rid of duplicate values is an important step in data analysis. There are lots of convoluted ways to do it, but Pandas has a super simple function you can use. Here is the  full documentation  for the function.                         ...
Read more

Why does DATA need a massage?

-
Image
  A bird-eye overview of the basic principles of data transformation: This post is a part of the “Analytics for startup” series so it might be helpful to the ones who just started figuring out how to build their own BI solution. I tried to answer the most common question that I frequently hear at the beginning of the BI implementation stage. Why do we need to do any data transformation? A very fair question having the fact that the data transformation can be the most challenging and the most time and money consuming part of the implementation. This is not any kind of tutorial but a bird-eye overview of the basic principles. Why do we need to transform the data? We don’t need to, actually. If the source is just one simple system or a couple of google sheets, then we don’t. In reality, though, we normally have many sources from the very beginning, e.g. some kind of CRM, Google analytics, google sheet, a database of our product, a website, etc. The main ideas of the transformation are...
Read more