Introducing advantage analytics using Power BI
A complete guide for advantage analytics using Power BI
One of the challenges for BI solutions in an organization to scale up is building a dashboard for every department. As a result, we build countless dashboards with very segmented pieces of information from different sources, and it's hard to see the root cause when problems arose. In this topic, I try to solve this problem within the ready BI solution for almost organization, the PBI. I will try to write another post on this problem using Tableau or Qlik in the next post. For now, you can download an example for this post here on Github.
- Data:
I'm using the sample superstore data that widely available for download. You can find the link here - Power BI:
Make sure you have ready installed some of the analytics function of PBI such as Key Influencers, Decomposition Tree or Forecasting with Arima
I) Explain the increase/decrease
Explain the increase is a tool that makes every chart self-explanatory, you don't need to look around to find the cause of increment or decrement.
Whenever you clicked on the chart → Analyze → Explain the increase. We can find out what element contribute to the changed. In the example above, we can see that the South and East regions accounted for the majority of the decrease among regions while the west accounted for the increment.
The second element, in terms of customers, Richard and Darrin accounted for the majority of the decrease among customers.
The factors are sorted by their weight, which means that the most important factor will show at the top. You can read the top description for more detail.
We can also change the visualization chart by click at the bottom (some charts type: Scatter Chart, Waterfall, Stack column, Ribbon).
II) Decomposition Tree
A decomposition tree is a tool that automatically aggregates your data, so you can easily see the contributing factors of each metric.
We can see that technology account for the majority of Profit. While Furniture still bring the profit but some Sub Category are losing money such as Bookcases and Tables with the shipping method of Standard Class. Tennessee accounted for the major loss. Read more about Decomposition tree here
III) Key Influencers
Key Influencer is a function that helps you understand which factors make your metrics change. It’s consist of 2 tabs:
- Key Influencers
- Linear Regression for continuous data
- Logistic Regression for Categorical data - Top Segments
- Decision tree
linear regression method (for continuous data), that shows you how the other factors affect your result. Examples below
- Key influencers on continuous data
As you the chart above self-explained that when Quantity goes up 173,48 the average metrics (here we choose Profit) increased by 1,38k which quite obvious. So to make this chart bring out the most valuable insight, we need to carefully select the right features for the inputting.
2. Key influencers on categorical data
As shown above, there are 10,77x more likely for the profit to be hight when Sub-category is Copiers flowing by State Marland. The column chart on the right-hand side compares the Subcategory copier with others. Copier account for 80% of the variance while the average variance is 8%
3. Segments of Key Influencers
We found 2 segments by % of the profit. Detail as below. we can see that when a discount is less or equal to 0 then profit is more likely to be high. So on so forth
Read more about Key Influencers here
IV) Predict times series data with Arima method
PBI provider build-in function to forecast times series data with R called forecasting with ARIMA. This allows you to have quick forecasting of your metrics with ease and adjustability.
To understand more about the ARIMA method you can refer to this post. But make sure you install R to your PBI. (Instruction here)
Above I have shown you some of the ready function within Power BI to deeply analyzing your data without having to look at many charts. Please leave comments if you have any questions. I really appreciated that
