<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<p>On 6/29/17 10:21 AM, G. Allegri wrote:<br>
</p>
<blockquote type="cite"
cite="mid:CAB4g1=xZEmkm53U-3dM8KCH1ta_bm8smw7JBf-+FoEp==aZYvA@mail.gmail.com">
<div dir="auto">
<div class="gmail_extra" dir="auto">
<div class="gmail_quote"><br>
<blockquote class="quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
1. Allowing multiple field selection for grouping<br>
2. Keeping attributes of first feature when grouping<br>
</blockquote>
</div>
</div>
<div dir="auto"><br>
</div>
<div dir="auto">I will accept this criteria, obviously, if it's
the preferred solution for the mosts. </div>
<div dir="auto">I just want to report that many users
(partecipants to courses or customers) say they find having
the "first feature value" misleading. </div>
<div dir="auto">I agree with them. I would set the field values
only for the grouping fields, having the same value within the
group, and set null for the other fields.</div>
<div dir="auto"><br>
</div>
<div dir="auto">The problem raises during a long workflow. At
some point you obtain a dataset with unconsisten field values,
and it's not always obvious to know when and.which field value
was set miningless.</div>
</div>
</blockquote>
<br>
In the long run the most flexible solution will be to allow
specifying the aggregate function applied to each field individually
("first/any", "mean", "mode", "max"...).<br>
<br>
No worries if it's not implemented right now, but please keep the
code modular enough to introduce this easily later down the road.
Without manually editing each algorithm that supports grouping.<br>
<br>
A possible approach to this would be a parameter
"ParamterFieldAggregation" that offers a standard gui to configure
the aggregation behavior (fieldA: min; fieldB: mean; fieldC: any;
fieldD: mean(expression("numBirds2017 - numBirds2000")) ). <br>
<br>
Matthias<br>
</body>
</html>