Neural Network Validation

bkruger

Hi,

I have a (basic - I think) question. When I do a NN model and performance validation, I can clearly see that 1 case was incorrectly classified (refer below). (There were more, but ignore that for now).

true CC true CF true GA class precision
pred. CC 1 0 2 33.33%
pred. CF 0 2 0 100.00%
pred. GA 1 0 11 91.67%
class recall 50.00% 100.00% 84.62%

This data is about authorship attribution, so I know that the author CC was mistakingly identified as author GA. So now I want to look at this specific record and investigate it further. How do I identify this specific case?

Any help would be appreciated.

Thanks
BK

SebastianLoh

Hi BK,

if I understood you right, you want to know which examples are classified wrong, is this correct?

In RM this is very easy. The "Filter Examples" operator allows you to filter wrong labeled examples. Take a look at the example process below to see how it works.

Ciao Sebastian

<宏/ >

<操作符法ivated="true" class="process" expanded="true" name="Process">

<操作符法ivated="true" class="retrieve" expanded="true" height="60" name="Retrieve (2)" width="90" x="45" y="210">


<操作符法ivated="true" class="retrieve" expanded="true" height="60" name="Retrieve (3)" width="90" x="15" y="89">


<操作符法ivated="true" class="decision_tree" expanded="true" height="76" name="Decision Tree" width="90" x="179" y="30"/>
<操作符法ivated="true" class="apply_model" expanded="true" height="76" name="Apply Model" width="90" x="246" y="210">


<操作符法ivated="true" class="multiply" expanded="true" height="94" name="Multiply" width="90" x="380" y="210"/>
<操作符法ivated="true" class="filter_examples" expanded="true" height="76" name="Filter Examples" width="90" x="581" y="255">


<操作符法ivated="true" class="performance" expanded="true" height="76" name="Performance" width="90" x="447" y="30"/>

bkruger

Thank you Sebastian, this is 100% what I was looking for.

Cheers BK