Weka从命令行预测CSV [英] Weka predictions to CSV from command line
本文介绍了Weka从命令行预测CSV的处理方法,对大家解决问题具有一定的参考价值,需要的朋友们下面随着小编来一起学习吧!
问题描述
这与以下问题类似:我们预测为CSV ,但从命令行
This is similar to this question: Weka Predictions to CSV, but from the command line.
我有以下Weka命令:
I have the following Weka command:
java -Xmx10G weka.classifiers.meta.FilteredClassifier \
-t test_data.arff -d prediction.model -p first -no-cv \
-F "weka.filters.unsupervised.attribute.Remove -R 1" \
-W hr.irb.fastRandomForest.FastRandomForest \
-- -I 512 -K 0 -S 512
其中给出以下数据:
=== Predictions on training data ===
inst# actual predicted error prediction (primary_key)
1 1:0 1:0 0.996 (r153)
2 1:0 1:0 0.994 (r756)
3 1:0 1:0 0.97 (r23)
4 1:0 1:0 0.995 (r18153)
5 1:0 1:0 0.947 (r2691)
这是非常好的,但我想将此数据输出到CSV文件,以便轻松加载到数据库中。我如何告诉Weka我想要CSV输出我的预测?
This is great, but I would like to output this data to a CSV file for easy loading into a database. How do I tell Weka that I would like a CSV export of my predictions?
注意:使用Weka 3.6.6版
Note: using Weka version 3.6.6
推荐答案
这对我有用:
java -Xmx28G weka.Run -no-scan weka.classifiers.meta.FilteredClassifier \
-classifications "weka.classifiers.evaluation.output.prediction.CSV -p first" \
-l $model \
-T $dataset \
|tail -n+6 |head -n -1
#note: |tail -n+6 <-- removes the header of file
# |head -n -1 <-- removes the footer of file
我必须升级到Weka 3.7.9才能使其工作。
I had to upgrade to Weka 3.7.9 to get this to work.
这篇关于Weka从命令行预测CSV的文章就介绍到这了,希望我们推荐的答案对大家有所帮助,也希望大家多多支持IT屋!
查看全文