k近邻 5-fold cross validation.doc

下载文档

8
0
约 3页
2017-05-27 发布于河南
举报
版权申诉
保障服务

k近邻 5-fold cross validation.doc

1、本文档共3页，可阅读全部内容。
2、原创力文档（book118）网站文档一经付费（服务费），不意味着购买了该文档的版权，仅供个人/单位学习、研究之用，不得用于商业用途，未经授权，严禁复制、发行、汇编、翻译或者网络传播等，侵权必究。
3、本站所有内容均由合作方或网友上传，本站不对文档的完整性、权威性及其观点立场正确性做任何保证或承诺！文档内容仅供研究参考，付费前请自行鉴别。如您付费，意味着您自己接受本站规则且自行承担风险，本站不退款、不进行额外附加服务；查看《如何避免下载的几个坑》。如果您已付费下载过本站文档，您可以点击这里二次下载。
4、如文档侵犯商业秘密、侵犯著作权、侵犯人身权等，请点击“版权申诉”（推荐），也可以打举报电话：400-050-0827(电话支持时间：9:00-18:30)。

k近邻 5-fold cross validation

Finish a python code that is for running 5-fold cross validation.Nearest neighbor code is provided. #recall Xtr_rows is 50000*3072 matrix import numpy as py import matplotlib.pyplot as plt X1=Xtr_rows[:10000,:]# take first 10000 for validation Y1=Ytr[:10000] X11=Xtr_rows[10000:,:]# keep last 40000 for train Y11=Ytr[10000:] X2=Xtr_rows[10000:20000,:]# take 10000-20000 for validation Y2=Ytr[10000:20000] X21=Xtr_rows[:10000,:] X22=Xtr_rows[20000:,:] X23=np.row_stack((X21,X22)) Y21=Ytr[:10000] Y22=Ytr[20000:] Y23=np.row_stack((Y21,Y22))# keep first 10000 and 20000-50000 for train X3=Xtr_rows[20000:30000,:]#take 20000-30000 for validation Y3=Ytr[20000:30000] X31=Xtr_rows[:20000,:] X32=Xtr_rows[30000:,:] X33=np.row_stack((X31,X32)) Y31=Ytr[:20000] Y32=Ytr[30000:] Y33=np.row_stack((Y31,Y32))# keep first 20000 and 30000-50000 for train X4=Xtr_rows[30000:40000,:]#take 30000-40000 for validation Y4=Ytr[30000:40000] X41=Xtr_rows[:30000,:] X42=Xtr_rows[40000:,:] X43=np.row_stack((X41,X42)) Y41=Ytr[:30000] Y42=Ytr[40000:] Y43=np.row_stack((Y41,Y42))# keep first 30000 and 40000-50000 for train X5=Xtr_rows[40000:50000,:]#take 40000-50000 for validation Y5=Ytr[40000:50000] X51=Xtr_rows[:40000,:]# keep first 40000 for train Y51=Ytr[:40000] validation_accuracies=[] for k in [1,3,5,10,20,50,100]: nn=NearestNeighbor() nn.train(X11,Y11) Y1_predict=nn.predict(X1,k=k) acc=np.nean(Y1_predict==Y1) print accuracy:%f %(acc,) validation_accuracies.append((k,acc)) for k in [1,3,5,10,20,50,100]: nn=NearestNeighbor() nn.train(X23,Y23) Y2_predict=nn.predict(X2,k=k) acc=np.nean(Y2_predict==Y2) print accuracy:%f %(acc,) validation_accuracies.append((k,acc)) for k in [1,3,5,10,20,50,100]: nn=NearestNeighbor() nn.train(X33,Y33) Y3_predict=nn.predict(X3,k=k) acc=np.nean(Y3_predict==Y3) print accuracy:%f %(acc,) validation_accuracies.append((k,acc)) for k in [1,3,5,10,20,50,100]: nn=NearestNeighbor() nn.t