Compiling:
CHC now requires the following sources from the cvs repository:
mlj/src/chc
mlj/src/id3
mlj/src/nb
mlj/src/shared
mlj/src/c45
Compile CHC from the src directory with the command
listed below.
javac chc/*.java
Running:
Run chc from the src directory with the following
command
java chc.CHCDriver [datafile] [options]
Datafile:
The data file to be tested with CHC. ie. vote, anneal, iris.
CHC uses the datafile system {.names, .all, .data, .test, .train, .val}
Make sure to include the path for the datafile if it is in another
directory.
Options:
finalgeneration=# - CHC will terminate at the end of
that generation - 1.
populationsize=# - the maximum number of Hypothesisi
spawned each generation.
inducer=[inducer] - the inducer to be used when running chc.
Options are naive, id3, c4.5.
bitmask=[mask] - mask is a string of 7 bits. 1 indicating on;
0 indicatinf off;
a description of the options controlled by the
bitmask is below in the Printing Options section
of this file in the order they appear. they
correspond to the CHC Printing options of the
MLJ-Options.file.
Options can also be controlled from the MLJ-Options.file
under the CHC options section
finalgeneration #
populationsize #
inducer [inducer]
FitnessDistribution #,#,#
These are the same options as above. except FitnessDistribution
which is used to implement the new outerloop fitness equation.
The three numbers should all add to 1.0 and correspond to a, b, and
c in the fitness equation.
Log Options
CHC uses the WRAPPER loglevel in the MLJ-Options.file.
loglevel output
0 only basic errors
1 prints basic generational information.
2 print out the process of chc.
3 internal working of chc plus non basic errors
4 print whole population each generation
5-6 no change
7 debugging
Inducers now run on the INDUCER loglevel in the
MLJ-Options.file and the CatTestResults uses the CTR.
If you are getting unwanted results try setting these
loglevels to 0.
Printing Options
They are in the MLJ-Options file.
print_all_hypothesis: prints all the hypothesesi
created durring this run.
best_overall_hypotheses: prints the top ten hypotheses
average_fitness_per_generation: prints the average fitness per
generation
best_Spawned_Hypothesis_of_generation: prints the information of the
best individual spawned in each
generation
best_overall_Hypothesis_of_generation: will print the information of
the best overall Hypothesis to
this point in the data run
best_fitness_of_each_generation: I forgot. nothing right now
though. don't worry about it.
correct_vs_incorrect_instances: will print the correctly
classified instances vs the
incorrectly classified instances.
Note:
Cataclysmic mutation is automatically implemented when needed.
It is used to fill any empty spots in the new population.
没有合适的资源?快使用搜索试试~ 我知道了~
数据挖掘的java算法
共191个文件
java:160个
class:21个
file:2个
5星 · 超过95%的资源 需积分: 9 47 下载量 17 浏览量
2010-05-11
17:07:03
上传
评论
收藏 423KB RAR 举报
温馨提示
数据挖掘的java算法数据挖掘的java算法数据挖掘的java算法数据挖掘的java算法
资源推荐
资源详情
资源评论
收起资源包目录
数据挖掘的java算法 (191个子文件)
add 52B
Graph.class 27KB
InstanceList.class 22KB
BaseInducer.class 7KB
Node.class 4KB
GraphMap.class 4KB
Edge.class 3KB
LogOptions.class 3KB
DisplayPref.class 2KB
GraphMap$ArrayCmp.class 905B
GraphMap$ArrayOrd.class 621B
Face.class 566B
Graph$MapEdgeOrd2.class 560B
Graph$MapEdgeOrd1.class 560B
Graph$NodeSorter.class 475B
Graph$EdgeSorter.class 475B
GraphObject.class 462B
Graph$Orderer.class 448B
ParamTypes.class 421B
ASCIIPref.class 416B
SortFunction.class 177B
OrderingFunction.class 160B
MLJ-Options.file 503B
Systems.file 338B
Graph.java 102KB
TDDTInducer.java 99KB
InstanceList.java 82KB
CatTestResult.java 48KB
Entropy.java 43KB
FileSchema.java 42KB
NaiveBayesCat.java 39KB
InstanceReader.java 36KB
SplitScore.java 35KB
CatDist.java 32KB
GetEnv.java 30KB
Basics.java 29KB
BaseInducer.java 28KB
PerfEstDispatch.java 27KB
SearchInducer.java 24KB
DecisionTree.java 23KB
MLJArray.java 22KB
Instance.java 22KB
SASearch.java 21KB
ID3Inducer.java 21KB
CatGraph.java 21KB
Inducer.java 20KB
MLJ.java 20KB
PerfEstimator.java 19KB
Schema.java 19KB
Bootstrap.java 19KB
SplitAttr.java 18KB
CrossValidator.java 17KB
Env_Inducer.java 16KB
Categorizer.java 16KB
NominalAttrInfo.java 16KB
StatData.java 15KB
AttrInfo.java 14KB
CHC.java 14KB
ProjectInd.java 14KB
SearchDispatch.java 14KB
BFSearch.java 13KB
BagCounters.java 13KB
NodeCategorizer.java 13KB
StratifiedCV.java 12KB
Hypothesis.java 12KB
State.java 11KB
Node.java 11KB
RealAttrInfo.java 11KB
ErrorData.java 11KB
PerfEstState.java 10KB
RealAndLabelColumn.java 10KB
NaiveBayesInd.java 10KB
StateSpace.java 9KB
GraphMap.java 9KB
PerfData.java 9KB
Population.java 9KB
NullInducer.java 8KB
CVIncremental.java 8KB
Edge.java 8KB
DisplayPref.java 8KB
RDGCategorizer.java 8KB
MEnum.java 8KB
StateSpaceSearch.java 8KB
FSSInducer.java 8KB
HoldOut.java 7KB
LeafCategorizer.java 7KB
FileNames.java 7KB
DataDistributor.java 7KB
u_Inducer.java 6KB
RootedCatGraph.java 6KB
CompState.java 6KB
LogOptions.java 6KB
Farmer.java 6KB
Globals.java 6KB
Matrix.java 6KB
ThresholdCategorizer.java 5KB
ConstCategorizer.java 5KB
AttrValue.java 5KB
PerfEstInfo.java 5KB
HCSearch.java 5KB
共 191 条
- 1
- 2
资源评论
- tjiezps2015-12-11调试算法进行 测试用sas跑出来和java跑出了是否一致
- qinpu198912222012-07-15算法很全不错
- m304792012-06-17包含有6个算法,挺不错的。就是自己水平有限,需要花一段时间去好好的读程序。 这个下载是2积分的,与另一个java数据挖掘算法下载需0积分,资源是一样的。弄得我白花了2积分!把两个全下载了
丛晓男
- 粉丝: 98
- 资源: 17
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功