I am trying to solve a simple classification problem where the label has 12 different levels and need to classify each example into one of these 12. However, I want my output to look like refer the image:
http://i.stack.imgur.com/49USG.png
Here; assuming that I set a confidence threshold of 20%; I want my output to contain all the labels for each id which are above 20% and ordered (highest confidence first). If none of the labels are above 20%; then a default label.
More specifically, are there any existing operators in Rapidminer which could give such an output?
Whenever the Apply Model operator runs, it produces new special attributes corresponding to confidences for the individual values of the label attribute. So if the label has values one, two, three, three new attributes will be created confidence(one), confidence(two), confidence(three). It would be possible to use the Generate Attributes operator to work out some logic to decide how to really classify each example. It would also be possible to use the Apply Threshold operator (with Create Threshold) to do something similar. It's impossible to give any more guidance unless you post a representative example with data.
Related
Although I have been able to split one of the legends into two columns on my map the other one is proving problematic in doing so. I cannot see any differences in the properties of each legend in terms of the item column customizing.
First of all, convert the legend to graphics (by right clicking the image and selecting convert to graphics). Follow this up by ungrouping the image into different elements. You can then just move the items across into a second column manually.
I would like to put together a more publication-worthy (and more easily produced) version of the diagram that I've hacked together below:
Basically, it's a two part illustration with an HTML table on the left (showing different cluster assignments from varying parameters) and a modified dengrogram plot on the right.
I'm guessing it's apparent from the snapshot what I'm trying to add to the plot… Basically it could be a set of horizontal, stacked bars that show how the different parameters grouped the observations into different clusters. I used brackets with the cluster numbers when drawing these by hand, but anything that underlines the right observation numbers would be acceptable.
I realize there are two parts to this question: How to get the two-up layout when one of the panels is HTML instead of a figure, and how to modify the plot. [If only one part gets answered here, I can ask for the other in a separate question.]
You can label your plot using line and text. You need to set the clipping property to off in order for lines to show up outside of your axes. Here's an example:
data = [1,3,4,5; 2,6,7,8; 9,3,7,4;3,8,5,2];
tree = linkage(data,'average');
figure()
dendrogram(tree)
set(gca,'Position',[0.13, 0.3,0.775, 0.65])
h1 = line([1,1],[4,4.8],'Color','k');
h2 = line([1,3],[4,4],'Color','k');
h3 = line([3,3],[4,4.8],'Color','k');
set(h1,'Clipping','off')
set(h2,'Clipping','off')
set(h3,'Clipping','off')
ht = text(1.7,3.9,'Label');
This is partially an academic question, but I think it can be done properly without javascript, and I like to keep javascript out of the handling of form data when possible.
Let's say I have single text input called "Name", and a table that has an arbitrary number of rows. Each row has two inputs in it, named "Color" and "Shape".
Let's say the text input is filled out as "Shape-Set-1" and there are 3 rows with the following values for Color and Shape:
"Blue" and "Square"
"Green" and "Circle"
"Yellow" and "Triangle"
When I post, the POST data will look like this:
Name: "Shape-Set-1"
Color: "Blue"
Color: "Green"
Color: "Yellow"
Shape: "Square"
Shape: "Circle"
Shape: "Triangle"
Great! All the data made it! But how do I know FOR SURE which Color and which Shape are connected? Are they always guaranteed to be gathered in the correct order? (My guess is no, as it probably depends on the browser, unless there is a standard for this). Is there a way to nest and have the rows report their Color and Shape inputs within an object?
Any ideas?
You would need to give each row an id, so that the inputs become Colour1 and Shape1, Colour2 and Shape2, etc, then you can parse the keys on the server side, determining the number to allow you to link the two together.
I dont know how forms go about serializing their data and passing it across, if it is guaranteed that inputs will appear in order then you know that the Colour array and Shape array index 0 will relate to the same item (I do not know if this is the case though, might be worth looking into).
In HTML forms input fields and their captions are declared as label and input tags, but which tags should be used for displaying the same data?
I'm considering either dl, dt, dd or label and disabled input tags, but both solutions don't feel right.
Of course there are many ways to "somehow" display the data, but I was wondering if there's a preferred way to do this.
Bear in mind that label/input is just a user-adjustable variation of the notion of key/value (or name/value) pairs. What you want is merely something that represents a key (label) and value (text) in such a way that the two can be distinguished from each other. The input performs the "value" function visually, and forces the user to regard the label as the name of the value. Without the visual clue of the input you can still make the name/value relationship obvious by bolding the name, putting it in a different color, different font, spacing, etc.
Even if you were to use <dd> and <dt> and the like, you would still need to make sure the name/value distinction was visually distinctive. (Especially if you were using a reset stylesheet that canceled out any inherent browser-generated difference between the tags.)
I think this depends on what the data is and how you'd describe it.
If you're showing a list of titles with associated reference data then a definition list seems to fit the bill. (Since it retains the semantic link between title and reference data).
This will be similar in how the label is directly linked to your form fields using the for attribute, making it not just about visual reference but also about data connection and relation.
I usually ask myself the question, "What am I trying to say here?" and "How do my data relate to one another".
Background info:
I was handed a "Tool", which was made using MS-Access 2007, and asked to add some things... The tool is basically a collection of options for querying a database. On a form titled CreatedReport there is a listbox that is bound to a table called analyzed which has all of resulting data from the query/queries that ran. The original creator of this tool set the column widths to specific values but with the new collection of possible results, those widths are very far off.
Desired Outcome:
The final result I want to achieve is, of course, to have the columns be the correct widths for the info that is in the columns. As long as that is achieved, I really don't care which route I have to take to get there.
Question:
How can I get the columns in a listbox in MS-Access 2007 to be sized appropriately for each use? Is there an auto-size feature I haven't stumbled across yet or do I need to hard code the set of column widths for each group? This wouldn't be too hard to do since there would only be about 4 or 5 different groups but I would prefer for the process to be automatic if at all possible.
Another approach would be to have the results returned in a sub form datasheet view, then the user can adjust the column widths also to set the widths automatically use code like this:
Example
This example takes effect in Datasheet view of the open Customers form. It sets the column to fit the size of the visible text.
Forms![Customers]![Address].ColumnWidth = -2
You could put this code into the Current Event of the sub form.
I don't think that Robert Harvey's answer is actually responsive to your question.
What you need to do is:
calculate the maximum length of the values in each column,
AND
figure out, based on the font in use, how wide the column should be.
Note that you may not actually want to set it to the maximum width if the value exceeds a certain threshold.
I don't know to do the second taks, but I suspect Stephen Lebans has already done the work on it. You might want to search his website for it.
Last time I checked, you still had to write code for this.
Your best bet is to use a resizer someone has already written. Here is a good one. It's old, but it shoul still work:
http://www.jamiessoftware.tk/resizeform/rf_jump.html
This is a quick solution that should help when you want to set up listview columns of different widths, and you know in advance the widths you want (eg, you know that column X will always be a 2-character State abbreviation, and column Y will always be a city name).
Just supply all the widths as a single semi colon-delimited string. Code each width as a number and a unit, such as 'in' or 'cm'. This worked well for me: Me.lsvPayHist.ColumnWidths = "1.0 in;0.8 in;1.0 in;1.0 in;2.0 in"