Mike de Waard:Machine Learning for Developers using Scala

相似文件 换一批

是 否 +2 论坛币

k人 参与回答

经管之家送您一份

应届毕业生专属福利!

求职就业群

赵安豆老师微信：zhaoandou666

经管之家联合CDA

送您一个全额奖学金名额~ !

立即领取

感谢您参与论坛问题回答

经管之家送您两个论坛币！

+2 论坛币

本帖隐藏的内容

Machine Learning for Developers Mike de Waard.pdf (14.29 MB, 需要: 1 个论坛币)

2016-4-21 09:42:27 上传

需要: 1 个论坛币  [购买]

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

分享0 收藏0 回帖

关键词：Developers developer Learning earning machine

1. object KNNExample {
   
2.    def main(args: Array[String]): Unit = {
   
3.     val basePath = "/.../KNN_Example_1.csv"
   
4.     val testData = getDataFromCSV(new File(basePath))
   
5.     }
   
6. 
   
7.   def getDataFromCSV(file: File): (Array[Array[Double]], Array[Int]) = {
   
8.     val source = scala.io.Source.fromFile(file)
   
9.     val data = source
   
10.         .getLines()
    
11.         .drop(1)
    
12.         .map(x => getDataFromString(x))
    
13.         .toArray
    
14. 
    
15.     source.close()
    
16.     val dataPoints = data.map(x => x._1)
    
17.     val classifierArray = data.map(x => x._2)
    
18.     return (dataPoints, classifierArray)        
    
19.   }
    
20. 
    
21.   def getDataFromString(dataString: String): (Array[Double], Int) = {
    
22. 
    
23.     //Split the comma separated value string into an array of strings
    
24.     val dataArray: Array[String] = dataString.split(',')
    
25. 
    
26.     //Extract the values from the strings
    
27.     val xCoordinate: Double = dataArray(0).toDouble
    
28.     val yCoordinate: Double = dataArray(1).toDouble
    
29.     val classifier: Int = dataArray(2).toInt
    
30. 
    
31.     //And return the result in a format that can later
    
32.     //easily be used to feed to Smile
    
33.     return (Array(xCoordinate, yCoordinate), classifier)
    
34.   }
    
35. }

复制代码

1. object KNNExample extends SimpleSwingApplication {
   
2.   def top = new MainFrame {
   
3.     title = "KNN Example"
   
4.     val basePath = "/.../KNN_Example_1.csv"
   
5. 
   
6.     val testData = getDataFromCSV(new File(basePath))
   
7. 
   
8.     val plot = ScatterPlot.plot(testData._1,
   
9.                                  testData._2,
   
10.                                  '@',
    
11.                                  Array(Color.red, Color.blue)
    
12.                                 )
    
13.     peer.setContentPane(plot)
    
14.     size = new Dimension(400, 400)
    
15. 
    
16.   }
    
17.   ...

复制代码

1. def main(args: Array[String]): Unit = {
   
2.    val basePath = "/.../KNN_Example_1.csv"
   
3.     val testData = getDataFromCSV(new File(basePath))
   
4. 
   
5.     //Define the amount of rounds, in our case 2 and
   
6.     //initialise the cross validation
   
7.     val cv = new CrossValidation(testData._2.length, validationRounds)
   
8. 
   
9.     val testDataWithIndices = (testData
   
10.                                 ._1
    
11.                                 .zipWithIndex,
    
12.                                 testData
    
13.                                 ._2
    
14.                                 .zipWithIndex)
    
15. 
    
16.     val trainingDPSets = cv.train
    
17.       .map(indexList => indexList
    
18.       .map(index => testDataWithIndices
    
19.       ._1.collectFirst { case (dp, `index`) => dp}.get))
    
20. 
    
21.     val trainingClassifierSets = cv.train
    
22.       .map(indexList => indexList
    
23.       .map(index => testDataWithIndices
    
24.       ._2.collectFirst { case (dp, `index`) => dp}.get))
    
25. 
    
26.     val testingDPSets = cv.test
    
27.       .map(indexList => indexList
    
28.       .map(index => testDataWithIndices
    
29.       ._1.collectFirst { case (dp, `index`) => dp}.get))
    
30. 
    
31.     val testingClassifierSets = cv.test
    
32.       .map(indexList => indexList
    
33.       .map(index => testDataWithIndices
    
34.       ._2.collectFirst { case (dp, `index`) => dp}.get))
    
35. 
    
36. 
    
37.     val validationRoundRecords = trainingDPSets
    
38.       .zipWithIndex.map(x => (  x._1,           
    
39.                                 trainingClassifierSets(x._2),
    
40.                                 testingDPSets(x._2),
    
41.                                 testingClassifierSets(x._2)
    
42.                                )
    
43.                         )
    
44. 
    
45.     validationRoundRecords
    
46.         .foreach { record =>
    
47. 
    
48.       val knn = KNN.learn(record._1, record._2, 3)
    
49. 
    
50.       //And for each test data point make a prediction with the model
    
51.       val predictions = record
    
52.         ._3
    
53.         .map(x => knn.predict(x))
    
54.         .zipWithIndex
    
55. 
    
56.       //Finally evaluate the predictions as correct or incorrect
    
57.       //and count the amount of wrongly classified data points.
    
58. 
    
59.       val error = predictions
    
60.         .map(x => if (x._1 != record._4(x._2)) 1 else 0)
    
61.         .sum
    
62. 
    
63.       println("False prediction rate: " + error / predictions.length * 100 + "%")
    
64.     }
    
65.   }

复制代码

1. val knn = KNN.learn(record._1, record._2, 3)
   
2. val unknownDataPoint = Array(5.3, 4.3)
   
3. val result = knn.predict(unknownDatapoint)
   
4. if (result == 0)
   
5. {
   
6.     println("Internet Service Provider Alpha")
   
7. }
   
8. else if (result == 1)
   
9. {
   
10.     println("Internet Service Provider Beta")
    
11. }
    
12. else
    
13. {
    
14.     println("Unexpected prediction")
    
15. }

复制代码

1. def getMessage(file : File)  : String  =
   
2.   {
   
3.     //Note that the encoding of the example files is latin1,
   
4.     // thus this should be passed to the fromFile method.
   
5.     val source = scala.io.Source.fromFile(file)("latin1")
   
6.     val lines = source.getLines mkString "\n"
   
7.     source.close()
   
8.     //Find the first line break in the email,
   
9.     //as this indicates the message body
   
10.     val firstLineBreak = lines.indexOf("\n\n")
    
11.     //Return the message body filtered by only text from a-z and to lower case
    
12. 
    
13.     return lines
    
14.         .substring(firstLineBreak)
    
15.         .replace("\n"," ")
    
16.         .replaceAll("[^a-zA-Z ]","")
    
17.         .toLowerCase()
    
18.   }

复制代码

1. def getFilesFromDir(path: String):List[File] = {
   
2.     val d = new File(path)
   
3.     if (d.exists && d.isDirectory) {
   
4.       //Remove the mac os basic storage file,
   
5.       //and alternatively for unix systems "cmds"
   
6. 
   
7.       d .listFiles
   
8.         .filter(x => x  .isFile &&
   
9.                      !x .toString
   
10.                         .contains(".DS_Store")  &&
    
11.                      !x .toString
    
12.                         .contains("cmds"))
    
13.                         .toList
    
14.     }
    
15.     else {
    
16.       List[File]()
    
17.     }
    
18.   }

复制代码

1. def main(args: Array[String]): Unit = {
   
2.     val basePath = "/Users/../Downloads/data"
   
3.     val spamPath = basePath + "/spam"
   
4.     val spam2Path = basePath + "/spam_2"
   
5.     val easyHamPath = basePath + "/easy_ham"
   
6.     val easyHam2Path = basePath + "/easy_ham_2"
   
7. 
   
8.     val amountOfSamplesPerSet = 500
   
9.     val amountOfFeaturesToTake = 100
   
10.     //First get a subset of the filenames for the spam
    
11.     // sample set (500 is the complete set in this case)
    
12.     val listOfSpamFiles =   getFilesFromDir(spamPath)
    
13.         .take(amountOfSamplesPerSet)
    
14.     //Then get the messages that are contained in these files
    
15.     val spamMails = listOfSpamFiles.map(x => (x, getMessage(x)))
    
16. 
    
17.      //Get a subset of the filenames from the ham sample set
    
18.      // (note that in this case it is not necessary to randomly
    
19.      // sample as the emails are already randomly ordered)
    
20. 
    
21.     val listOfHamFiles =   getFilesFromDir(easyHamPath)
    
22.         .take(amountOfSamplesPerSet)
    
23.     //Get the messages that are contained in the ham files
    
24.     val hamMails  = listOfHamFiles
    
25.         .map{x => (x,getMessage(x)) }
    
26.   }

复制代码

......

别问我为什么就为了签个到