WEKA预测

Question

我希望有人能帮助我。我是Weka的初学者，我在编写Java程序时遇到了一些问题。

我的目标是预测下一个时期/季度的销售数字，同时将其连接到我收集的用户生成的内容。

所以我想用WEKA预报器？！

我有以下问题：

1.我应该使用哪些设置来分别确定模式属性的影响

2.有用的建议？

谢谢!

这是我生成的.arff文件：

@relation Sales_total_Apple

%Sales: total sold apple products in quarter yyyy-qq
%Volume: number of user generated content (e.g. review rating) in quarter yyyy-qq
%Reviewrating: average value of reviewrating (e.g. amazon stars) in quarter yyyy-qq
%Valence: avarage value of user generated content valence (used sentiment analysis) in yyyy-qq
%Date: should be year-quarter

@attribute Sales numeric
@attribute Volume numeric
@attribute REVIEWRATING numeric
@attribute VALENCE numeric
@attribute Date date 'yyyy-mm'

@data
22672000,133.0,3.7954545454545454,0.25172045454545455,2007-01
12066000,84.0,3.8674698795180724,0.19313253012048195,2007-02
11579000,48.0,3.7872340425531914,0.11405957446808514,2007-03
13483000,24.0,3.9565217391304346,0.2705391304347826,2007-04
26755000,202.0,4.039800995024875,0.09474179104477576,2008-01
14636000,227.0,4.243362831858407,0.1526902654867257,2008-02
13507000,202.0,4.039800995024875,0.1419736318407955,2008-03
20555000,226.0,4.288888888888889,0.11981422222222161,2008-04
29614000,467.0,4.008583690987124,-0.0031933476394850266,2009-01
17022000,376.0,4.1226666666666665,0.009081333333333313,2009-02
18026000,206.0,4.058536585365854,0.03965170731707281,2009-03
20597000,248.0,3.866396761133603,0.055700809716599034,2009-04
33069000,738.0,4.150610583446404,-0.061250881953867,2010-01
22580000,743.0,4.063342318059299,-0.0537954177897575,2010-02
24546000,475.0,4.090717299578059,-0.050239240506329344,2010-03
31226000,491.0,4.189795918367347,-0.062333469387755164,2010-04
47146000,848.0,4.128689492325856,0.08620141676505302,2011-01
36118000,1201.0,4.31,0.030831249999999866,2011-02
41066000,512.0,4.039138943248532,-0.026479256360078324,2011-03
39910000,647.0,4.139318885448916,-0.02018513931888557,2011-04
73073000,1463.0,4.272913816689466,0.17992134062927442,2012-01
58552000,2038.0,4.3460972017673045,0.1778528718703978,2012-02
53841000,1808.0,4.348644161593802,0.1730128942999446,2012-03
51213000,1570.0,4.293180369662205,0.19067469725940053,2012-04
87389000,1463.0,4.272913816689466,0.17992134062927442,2013-01
66492000,1484.0,4.226567768037762,0.16101463250168524,2013-02
54181000,2012.0,4.193436101442068,0.16231909497762292,2013-03
55948000,867.0,4.094688221709007,0.21768198614318673,2013-04

Answer 1

如果你使用的是3.7.11或更高版本，你可以从包管理器安装时间序列包(见附件)。

​

​

使用它时，请转到预测选项卡，然后转到高级配置子选项卡。您可能希望首先尝试lag创建来实现ARMA模型。

​

​