Guest User

Untitled

a guest
Aug 19th, 2019
66
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
  1. import numpy as np
  2. import pandas as pd
  3. import time
  4. from sklearn.ensemble import RandomForestClassifier
  5. from sklearn.externals import joblib
  6. import scipy.io as sio #存mat
  7.  
  8. start=time.clock()#开始计时
  9.  
  10. # x_4=joblib.load('reduced_x_4.pkl') #读取数据
  11. # x_5=joblib.load('reduced_x_5.pkl')
  12. x_4=joblib.load('x_4.pkl') #读取数据
  13. x_5=joblib.load('x_5.pkl')
  14. x_6=joblib.load('x_6.pkl')
  15. x_7=joblib.load('x_7.pkl')
  16. y_4=joblib.load('y_4.pkl')
  17. y_5=joblib.load('y_5.pkl')
  18. y_6=joblib.load('y_6.pkl')
  19.  
  20. xtrain=x_6
  21. ytrain=y_6[:,1]
  22. xtest=x_7
  23. # ytest=y_5[:,1]
  24.  
  25. #RF
  26. clf=RandomForestClassifier(n_estimators=50,max_depth=25)
  27. clf.fit(xtrain,ytrain) #.astype('float')强制把类型定义为float,不然dtype=object会报错
  28. prediction1=clf.predict_proba(xtest)
  29. prediction=np.zeros((prediction1.shape[0],1))
  30. for i in range(0,prediction1.shape[0]):
  31. if prediction1[i,1]>0.5:
  32. prediction[i,0]=1
  33. else:
  34. prediction[i,0]=0
  35.  
  36. # 准确率P和召回率R。真正例TP,假正例FP,真反例TN,假反例FN
  37. # TP,FP,FN=0,0,0
  38. # for i in range(len(prediction)):
  39. # if ytest[i]==1 and prediction[i]==1:
  40. # TP+=1
  41. # if ytest[i]==0 and prediction[i]==1:
  42. # FP+=1
  43. # if ytest[i]==1 and prediction[i]==0:
  44. # FN+=1
  45. # P=TP/(TP+FP)
  46. # R=TP/(TP+FN)
  47.  
  48. index=np.lexsort([-1*prediction1[:,1]]) #按第二列降序排序
  49. prediction2=prediction1[index,:]
  50. print("Time used:",time.clock()-start,'s')#计算时间
  51. # print('P=',P)
  52. # print('R=',R)
  53.  
  54. # '''存mat('文件名.mat', {'文件里变量名': 要存的变量名})'''
  55. # sio.savemat('x_4.mat', {'x_4': x_4})
  56. # sio.savemat('x_5.mat', {'x_5': x_5})
  57. # sio.savemat('y_4.mat', {'y_4': y_4})
  58. # sio.savemat('y_5.mat', {'y_5': y_5})
  59.  
  60. # np.savetxt("prediction_7.csv", prediction, delimiter = ',')
RAW Paste Data