Pages: [1]
  Print  
Author Topic: Confidence and Prediction all the same  (Read 675 times)
nelze
Newbie
*
Posts: 18


« on: March 05, 2014, 04:22:50 AM »

Good day! I'm trying to read an excel file (using the Read Excel operator) in RapidMiner, but when I look at the results, the confidence and prediction are all the same values. However, when I use a CSV file (using the Read CSV operator) as the unlabeled data, it seems to predict it properly. The CSV file and the EXCEL file are both the same data, just in different formats, so I don't understand why the Excel version would give strange results. Could I have missed a setting or something? Any help will be appreciated.

Thank you very much!
« Last Edit: March 05, 2014, 04:37:51 AM by nelze » Logged
Marco Boeck
Administrator
Hero Member
*****
Posts: 953


WWW
« Reply #1 on: March 05, 2014, 09:20:03 AM »

Hi,

to see whether your setup of Read Excel and Read CSV produces identical output, you could store the output of each one (using the Store operator) and then open both stored example sets and compare them.

Regards,
Marco
Logged

nelze
Newbie
*
Posts: 18


« Reply #2 on: March 05, 2014, 09:43:53 AM »

Oh I'm sorry, what I mean by prediction being the same, I mean this:

Correct Prediction (using CSV), I have 4 confidence values because I am predicting 4 classes. The last column is the prediction:

0.8860144466097823   0.0   0.11398555339021778   3.0815034919635467E-19    1
0.6143980539313239   0.0   0.38560110537192616   8.406967499335851E-7    1
0.9859569967284759   0.0   0.014043003271520669   3.459433208878122E-15    1
0.4719509186438967   0.0   0.5270853842058498   9.636971502535191E-4    3
0.9837738580587952   0.0   0.01622614194120045   4.405660458971408E-15    1
0.9967563332443505   0.0   0.0032436667556495755   5.2553155952136636E-23    1
0.5759288037256314   0.0   0.42400890432723454   6.229194713409175E-5    1
0.9787821737511498   0.0   0.02121668285668769   1.1433921625919423E-6    1
0.9996751702426111   0.0   3.248297573889029E-4   1.4578020956211202E-18    1
0.8411716684610979   0.0   0.15878491911697307   4.341242192905734E-5    1
0.17253194445660552   0.0   0.1644529101772081   0.6630151453661863   4
0.18095747514120386   0.0   0.08592915193222532   0.7331133729265707   4

When I use an excel file (same file as CSV, just changed format. As in I just opened the same CSV file with the unlabeled data and save as excel):

1 0 0 0 1
1 0 0 0 1
1 0 0 0 1
1 0 0 0 1
1 0 0 0 1
.....

All the other rows have the same confidence and prediction, which is very strange.
« Last Edit: March 05, 2014, 09:48:50 AM by nelze » Logged
nelze
Newbie
*
Posts: 18


« Reply #3 on: March 05, 2014, 12:34:57 PM »

I also get this in the log

The value types between training and application differ for attribute 'K', training: real, application: integer

What can I do with that? Sometimes the number is a decimal but sometimes it is an integer.
Logged
Pages: [1]
  Print  
 
Jump to: