Tools I’ll be using for the Demo:
SQL Server 2012 (specifically SQL Server Analysis Services)
Sample data-set that comes with the excel add-in
Marketing Department needs to create Targeted Mailing list.
What data do we need?
To create a Targeted mailing list – we’ll need a historical data-set of customer purchase history
What will we do with the data?
Based on the historical data-set, we’ll be able to find “patterns” in the past consumer behavior. E.g. A single male going to college living in Europe is likely to buy a bike. And the using these patterns – we would then classify NEW customers.
Technically, we’ll be using the classification method using the Microsoft’s decision Tree algorithm
(Read the difference between classification and clustering)
Let’s get in action!
STEP 1: Build a Model
Data Mining Tab > click on classify:
Follow the steps:
Select the data:
In this case, since we want to predict the likelihood of buying a bike – our column to analyze is BikeBuyer
For the Demo, I am going to just leave it default. There are “optimization” steps that you can do but for the demo I am going to keep it super simple
Name the model:
The Model has been created!
STEP 2: Query the MODEL to predict the likelihood of bike purchase of a new customer
Select the model:
Select the data:
Specify the columns that would be used in predicting the likelihood:
Add the column that will have the “predicted value”
And example of Data Mining Expressions (DMX):
For the demo, I am just going to add the column to the existing table:
Yay! Here’s our Targeted Mailing list – see the last column:
Marketers can now send “coupons” to ONLY those people who are most likely to buy a bike! And so that’s how you create a targeted mailing list using the Excel Data Mining add-in.
- Excel data Mining in Action: Forecasting Twitter Followers for next week (parasdoshi.com)
- Machine Learning VS. Data Mining (parasdoshi.com)