I am new to WEKA and want to get the probability distribution of all instances over all classes from classify tab to a csv file for further analysis of models.
Please suggest how to go about that in WEKA GUI explorer?
I just noticed that there is a more options button in test options under the classify tab. There you can choose CSV for output predictions. If you click on CSV you can turn outputDistribution to true and name your file.Hope this helps.
If this doesn't work then its fairly easy to do this with java code.
Not completely sure what you are asking but I gather it is the output from a cross validation/% split run that you want as a csv file? If so, the best place to achieve this is in the Experimenter tab in WEKA. First, click 'New' and add your datasets in the bottom left box after clicking 'add new.' Secondly, add your algorithms in the bottom right box having clicked 'add new.' Finally, change your results destination to 'CSV file' and browse for the file location. You can configure the experiment type and number of repetitions here also. Run the experiment in the next tab and wait for completion. You will find your results in the destination csv.
Related
I have tables in excel and I'd like to copy/paste them into draw.io table. Is there any way I can make all rows into individual rows (texts) in draw.io instead of copy/paste each row (text) from excel to draw.io?
Thanks in advance!
Click the Plus + icon on the toolbar above the canvas.
Go to Advanced > CSV
There Just after the First 4 lines of comments (starting with '##'),
Replace everything, and add your list items. Click Import.
For learning how to change further styles, configs etc., read the help items in the text that was written by default in the box (which you replaced from 4th line). Or see here https://drawio-app.com/import-from-csv-to-drawio/.
Basically it is not something fully supported but in some cases copy/paste might work but use keyboard shortcuts.
Other possible solutions are to try to export as HTML/CSV and import it to draw.io
Yes, you can easily paste a part of table from microsoft excel and google sheet to
draw.io by 'crtl+c, crtl+v' creating a new table. It will be created with style "text;whiteSpace=wrap;html=1;" and then can be edited as a natively created draw.io table with actions like "create new column left" and so on.
You can use sql scripts for this.
-> Advance -> SQL .
I created sql scripts with excel formula and paste them here.
also please make sure you place the cursor properly , so that tables dont get overlapped.
I'm using REDMINE.
I have a list of projects on the screen.
Is it possible to add details of the ticket content to this screen?
I'd like to look over the screen at once instead of clicking with the mouse, how do I do that?
Also, if you have any plugins or useful tips and so on, please let me know.
Tamazawa, I believe you might be referring to "My Page" of Redmine where you could use widgets to decide what you see on that dashboard - keep in mind that such dashboard s project independent, so it can display issues from many projects that you are a member of on 1 screen.
For instance from My Page, you should be able to select an "Issues" widget to Add and set it up to pull a list of items to display using a saved custom query.
That should give you a snapshot of tasks at hand, without the need to click around the screen too much.
You may need to set up your saved query such that:
Description and / or
Last notes
are checked (if you want to see issues at a glance, without the need to open each one separately).
Alternatively, you can skip that "My Page" idea altogether, if you're ok with just opening a saved query with Description and / or Last notes selected.
I'm new to powercenter and the learning curve of this tool seems to be different from the programming languages I'm used to. Where I work everyone tells me to ask people in the office. Access to the internet is non existant. Stack overflow is one of the few places where I can look for info.
I would like to do something basic like, in the navigator window, there are sometimes thousands of mappings with veeeery similar names, I would like to know if there is a way to enable a "control-f" type of functionality. As of right now, placing the mouse on the window and writing the name kinda helps, but since the names are stuff like
k_l_ll_sigpir_proyectname_mappingname
and there are dozens and hundreds that share the same structure: "k_l_ll_sigpir_" trying to navigate when you don't know where info is other than the name is a bit of a nightmare, compared to other tools at least.
Can someone point me in the right direction? Thanks in advance.
I guess I was able to find something that might help, although this is a bit counter-intuitive. I'm assuming you want to find and open some mapping in Mapping Designer. Here goes:
After you've connected to Repository and have the right folder opened (not just connected!), choose Tools => Queries from the menu bar. A Query Browser should open for you.
Choose New to create a new query
In the Query Editor there's a grid. Choose Object Name in the Parameter Name column. Change the Operator to Contains and put whatever you're looking for as Value 1 (using your example: mappingname)
Click the blue play button on the menu bar or choose Query => Execute from top menu. Give this a few seconds.
Give it a few more seconds...
A Query Results window should be opened for you. NOTE: if you've been clicking anything it might get opened in the background - just check the Designer button on Start Bar. It should contain all the objects that contain mappingname string it the name. We've not limited this to mappings only though (I'll discuss this below).
=== now here's the odd part ===
You can right click any item on the list in Query Results window and choose Open in Workspace. And it won't work :)
Switch back to Query Editor window and close it (don't close the Query Results!). You may save you'r query if you like to.
You should be back at Query Browser. Close it.
Return to Query Results.
Right click the desired item and choose Open in Workspace.
Voila!
Now some remarks:
The whole repository is searched. The Query Results may therefore contain items from other folders, than the one you have open. Trying to Open in Workspace such objects won't work since the folder is not open in Designer.
You may construct more sophisticated queries. Just click 'Add new...' icon on the Query Editor window. This will add a line and allow you to choose extra conditions. It will be spawned by a AND condition which you may change to OR. You get the idea.
PS.
Yes, it's odd, I know. Just don't blame me ;)
The quick answer is that the 'metadata manager' tool can do exactly what you ask for. It is basically a 'google search' for your entire data warehouse. Only hiccup is that it requires a lot of work from the admins of your power center servers to get it up and running...
Alternatives:
Basically it depends on what you want to DO with the mapping after you've found it and why you look for it in the first place:
1) in the 'designer tool' you can look under 'targets' in the right hand tree structure
- when you've found the target table you want the mappings for
- simply right-click and choose 'dependencies'
- in the following dialog uncheck everything except 'mappings'
- in the following right click on the most likely mapping and choose 'open in workspace'
2) in the repository database there is a view called opb_task_inst_run that contains a row for each session that has ever run. You can write simple sql against that view with '%like patterns%' to locate mappings... you may want to add a where clause to 'start time' for the most recent week or so - especially if the repository is old and runtime metadata has never been deleted...
Please write back if you need something different...
I'm not aware of the functionality you're after within informatica, however, the mappings are listed in alphanumeric order which should alleviate this. Within the mapping itself there is search functionality to find transformations within the mapping.
As far as a place to start learning Informatica, informatica tv is one place to get your feet wet https://network.informatica.com/support-videos.jspa
There is an Address MVG field on Contact record. While creating new record if I click on this address MVG field, MVG applets takes a lot of time to get open (20 min). Kindly help to fix this. Or let me know the possible fixs or ways to fix it.
Thanks.
The first question that you should answer is: does it happen because of Siebel or the database?
To determine it, you can use a dedicated client with the SQL spool enabled, by adding the /s parameter to the command line in the shortcut. For example: D:\Siebel\Client\BIN\siebel.exe /c D:\Siebel\myConfig.cfg /s D:\Siebel\sql-spool.txt.
That way it will create a file with every SQL sentence being executed. This file will be quite big, so I suggest you to reproduce all the steps except the one causing trouble (open your Siebel client, navigate to the contacts screen, create a new record...), then open the SQL spool file in a text editor and position yourself at the end of the file. After that, click on the MVG field in Siebel, and refresh the spool file, to see what SQL sentences have been added.
The next step would be to launch the same sentences using a DB client (TOAD, Tora, SQL Plus...) and check if any of them takes too long to finish. You can access to the Oracle database administration console while doing so, and review the performance and the execution plan for each sentence, which can help you find what's causing the trouble. Most likely, you'll need to create an index for some columns.
If all the SQL sentences execute in a reasonable time, then you can discard the database and continue looking for problems in your SRF/repository:
Do you have custom code in any business component or applet involved (contacts, addresses, etc.).
Is your MVF field well defined? Is it vanilla or custom made?
Are there user properties which apply to the MVG fields?
Try disabling any customisation you've made and check if it still happens with vanilla Siebel.
My requirements are to create two buttons. One button that will download all of the data in an interactive report (Select * from DB_TABLE). The other button will download the interactive report data based on filter items (Select * from DB_TABLE where Name = :P1_Name)
I thought about utilizing javascript:gReport.controls.download(); but that didn't seem to work.
Any ideas or sites you can point me to? Many thanks in advance.
And no, I don't know why they want two separate buttons to download the report instead of using the 'Download' button in the interactive report menu.
Set the Request value of the button to CSV, so your link looks like f?p=181:9:3839908094418724:CSV::::
This will only perform the default CSV download though. There is however no built-in way to get the query from an IR. What i mean by that is that an IR which has filters applied.
If you only want to download results which are filtered through page items in the where clause, and don't bother with any IR filters or such, you could set up a report on another page, and have your link for a second button target that page (with CSV in request, but do use a redirect). The browser will prompt you with a save dialog, but you'll not end up on the page.