Key influencers

[shtak]

Hi all,

We're trying to use 'key influencers' visual object which quite a lot of experts

recommend as fantastic.

Our situation is to find factors that impact the sales, it doesn't seem to work alright.

The following explanation says this visual is for 'logistic regression' which is
'to estimates the probability of an event occurring, such as voted or didn't vote, based on a given data set of independent variables'
   https://www.ibm.com/topics/logistic-regression

reference on logistic regression
  https://addendanalytics.com/blog/exploring-power-bis-key-influencers

Now our question is if it is not appropriate to use this visual to 

find impacting factors for sales which is not binary values.

Any comment appreciated.

Regards,

Shinya

[shtak · Accepted solution]

Hi Albert,

Thank you very much for your kind and profound answer.
We have understood that key influencers visual is not suitable for sales factor analysis
while other statistical methods are available.

Regards,

Shinya

[v-heq-msft]

Hi @shtak ,
Based on your description, the Key Influencers visual object in Power BI is indeed a powerful tool for identifying the factors that have the greatest impact on binary outcomes, such as whether a customer will be lost or a product will be defective.
However, if you are trying to use this visual object to find factors that influence sales, which is a continuous variable, then this may not be the most appropriate tool. Logistic regression is designed for binary outcomes, not continuous outcomes like sales.
For finding factors that influence sales, you may want to consider using other statistical methods or machine learning algorithms that are suitable for regression tasks, such as linear regression, decision trees, or random forests. These algorithms can help you identify the factors that have the greatest impact on sales and provide more accurate predictions and insights.
You can check out these documents below which describe how to create linear regression models, decision trees and random forests using dax. It also describes how to differentiate between decision trees and random forests. You can choose one of them to use.

Decision Tree: Concepts- Part 1 - RADACAD
Implementing linear regression in Power BI - SQLBI
Random Forest: The Math of Intelligence (linkedin.com)
Decision trees vs random forests: a comparison for BI classification (linkedin.com)

Best regards,
Albert He

If this post helps, then please consider Accept it as the solution to help the other members find it more quickly