Database Reference
In-Depth Information
states: false positives, false negatives, true positives, and true negatives). The data table includes col-
umns for predicted Total Cost, predicted Total Profit, and Score. The values in these columns are cal-
culated using our input data and cost/profit factors.
Profit for Various Score Thresholds
This report section to the right of cost and profit inputs uses a line chart to show where the sug-
gested profit maximization threshold falls. The horizontal axis represents all of the possible scores
from the Score column in the data table (described in the previous section) and the vertical axis rep-
resents the Total Profit column from that same table.
Cumulative Misclassification Cost
for Various Score Thresholds
Cumulative Misclassification Cost for Various Score Thresholds section is located below the Profit for
Various Score Thresholds chart (refer to Figure 14-15) and shows two area segments to represent
both FPs in blue and FNs in red. This example only shows an area plot for FP because there isn't a cost
to be associated with FNs.
Shopping Basket Analysis
You're probably familiar with shopping basket algorithms from shopping online. Though not the only
weapon in the arsenal for Web storefront businesses, this approach to maximizing customer transac-
tions is simple and has been around for years. The idea is that historical data shows which items cus-
tomers tend to buy in bundles, and this information can be used in future transactions by suggesting
to you other items you should purchase. This targeting is relatively non-intrusive and can even be
seen as a helpful reminder to buy something useful (given the other items in the cart) that you would
have otherwise forgotten to purchase. From the business perspective, the targeting is a way to maxi-
mize the value of each transaction.
To see the Shopping Basket Analysis tool in action, follow these steps:
1.
In the
DMAddins_SampleData.xlsx
sample workbook, go to the worksheet called
Associate and click inside the Excel table.
2.
From the Analyze tab, click the Shopping Basket Analysis button.
3.
In the Shopping Basket Analysis dialog box, shown in Figure 14-16, choose the following
inputs:
Transaction ID:
This is where you identify a column that represents a transaction ID. This
column gives the association algorithm a way to group rows that belong to the same
transaction. This column could be called OrderID or something else, as long as it is the
identifier of a transaction.
●
Search WWH ::
Custom Search