Linear regression analysis helps us to predict the value of a variable based on the value of another variable. The variable to be predicted is known as the In this article, we will be creating a linear regression model that predicts the number of ice cream sales based on temperature using Java and GridDB. Let’s begin by importing the java libraries that will enable us to work with GridDB:<\/p>\n The dataset to be used shows the number of ice cream sales for different temperature values. It has been divided into two, that is, training and test datasets. A GridDB container can be seen as a SQL table.<\/p>\n For us to write data into the GridDB container, we should first connect to GridDB. We have to create a We have also created the We want to store the training dataset into GridDB. Let us first define the data rows as instances of the Let us now select the We can use the We now want to retrieve the training data from GridDB and use it to fit a machine learning model. We will write a TQL query to retrieve all the data in the We have used the We will use the Weka machine learning library to fit a linear regression model. Hence, we must convert our data into Weka instances. We will first create attributes for the dataset and store them in a FastVector data structure. We can then create Weka instances of the dataset.<\/p>\n Let’s first create the data structures for storing the attributes and the instances:<\/p>\n Next, we create a Let's use the Since the data instances are ready, we can fit a machine learning model. But first, let us import the necessary libraries from Weka:<\/p>\n Let us specify the class attribute for the dataset before feeding it into the model:<\/p>\n We can now use the Our linear regression model is now ready!<\/p>\n We will evaluate the model using the training dataset and test it using the test dataset. Let us define an array named For us to feed the data into the model, we must first convert it into Weka instances. We will store the attributes in a FastVector data structure and the instances in an ArrayList. Let us define them:<\/p>\n We can now use a Let us create instances of the training dataset and store them in an Instance variable named Let us specify the class attribute of the dataset before feeding it into the model:<\/p>\n Next, we feed the data into the model and print the evaluation summary:<\/p>\n We can now use our model to predict the number of ice cream sales for a particular temperature value. Let's use the last instance of the test dataset to make the prediction:<\/p>\n Download the Weka API from the following URL:<\/p>\n Set the class paths for the Note that the above commands may change depending on the location of the files.<\/p>\n Next, run the following command to compile your Execute the generated .class file using the following command:<\/p>\n The model returned a correlation coefficient of 0.9456. Correlation values range between -1 and 1, where 1 is very strong and linear correlation, -1 is inverse linear relation and 0 means no relation. The model also predicted 198 ice cream sales.<\/p>\n","protected":false},"excerpt":{"rendered":" Introduction Linear regression analysis helps us to predict the value of a variable based on the value of another variable. The variable to be predicted is known as the dependent variable while the variable used for predicting other variables is known as the independent variable. It estimates the coefficients of the linear equation with one […]<\/p>\n","protected":false},"author":41,"featured_media":30008,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[121],"tags":[],"class_list":["post-46793","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-blog"],"acf":[],"yoast_head":"\ndependent<\/code> variable while the variable used for predicting other variables is known as the independent variable<\/code>. It estimates the coefficients of the linear equation with one or more independent variables. Linear regression creates a straight line that minimizes the differences between the predicted and the expected output values.<\/p>\nTemperature<\/code> will be the indepedent variable while ice cream sales<\/code> will be the dependent variable.<\/p>\nImport Packages<\/h2>\n
import com.toshiba.mwcloud.gs.Collection;\nimport com.toshiba.mwcloud.gs.GSException;\nimport com.toshiba.mwcloud.gs.GridStore;\nimport com.toshiba.mwcloud.gs.GridStoreFactory;\nimport com.toshiba.mwcloud.gs.Query;\nimport com.toshiba.mwcloud.gs.RowKey;\nimport com.toshiba.mwcloud.gs.RowSet;\nimport java.util.*;\n\n\nimport java.util.Scanner;\nimport java.io.IOException;\nimport java.util.ArrayList;\nimport java.util.Properties;\nimport java.util.logging.Level;\nimport java.util.logging.Logger;<\/code><\/pre>\n<\/div>\nWrite the Data into GridDB<\/h2>\n
\nHowever, we want to move the training dataset into GridDB as it offers various benefits including faster query performance. We will define a GridDB container for storing the data. The container will be defined as a java class as shown below:<\/p>\n static class SalesData {\n @RowKey int id;\n double temperature;\n double icecreamsales;\n }<\/code><\/pre>\n<\/div>\nProperties<\/code> file from the java.util<\/code> package and provide the credentials of our GridDB installation using the key:value<\/code> pairs syntax as shown below:<\/p>\nProperties props = new Properties();\nprops.setProperty(\"notificationMember\", \"127.0.1.1:10001\");\nprops.setProperty(\"clusterName\", \"myCluster\");\nprops.setProperty(\"user\", \"admin\");\nprops.setProperty(\"password\", \"admin\");\nGridStore store = GridStoreFactory.getInstance().getGridStore(props);<\/code><\/pre>\n<\/div>\nstore<\/code> variable of type GridStore<\/code> to help us interact with the database.<\/p>\nStore the Training Dataset in GridDB<\/h2>\n
SalesData<\/code> class.<\/p>\nSalesData row1 = new SalesData();\nrow1.id=1;\nrow1.temperature=14.2;\nrow1.icecreamsales=215;\n\nSalesData row2 = new SalesData();\nrow2.id=2;\nrow2.temperature=16.4;\nrow2.icecreamsales=325;\n\nSalesData row3 = new SalesData();\nrow3.id=3;\nrow3.temperature=11.9;\nrow3.icecreamsales=185;\n\nSalesData row4 = new SalesData();\nrow4.id=4;\nrow4.temperature=15.2;\nrow4.icecreamsales=332;\n\nSalesData row5 = new SalesData();\nrow5.id=5;\nrow5.temperature=18.5;\nrow5.icecreamsales=406;\n\nSalesData row6 = new SalesData();\nrow6.id=6;\nrow6.temperature=19.4;\nrow6.icecreamsales=412;\n\nSalesData row7 = new SalesData();\nrow7.id=7;\nrow7.temperature=25.1;\nrow7.icecreamsales=614;\n\nSalesData row8 = new SalesData();\nrow8.id=8;\nrow8.temperature=23.4;\nrow8.icecreamsales=544;\n\nSalesData row9 = new SalesData();\nrow9.id=9;\nrow9.temperature=18.1;\nrow9.icecreamsales=421;\n\nSalesData row10 = new SalesData();\nrow10.id=10;\nrow10.temperature=22.6;\nrow10.icecreamsales=445;\n\nSalesData row11 = new SalesData();\nrow11.id=11;\nrow11.temperature=17.2;\nrow11.icecreamsales=408;<\/code><\/pre>\n<\/div>\nSalesData<\/code> GridDB container where the data will be stored:<\/p>\nCollectionput()<\/code> function to add the data into the container:<\/p>\nsd.put(row1);\nsd.put(row2);\nsd.put(row3);\nsd.put(row4);\nsd.put(row5);\nsd.put(row6);\nsd.put(row7);\nsd.put(row8);\nsd.put(row9);\nsd.put(row10);\nsd.put(row11);<\/code><\/pre>\n<\/div>\nRetrieve the Training Data<\/h2>\n
SalesData<\/code> container as shown below:<\/p>\nQueryselect *<\/code> TQL statement to retrieve all the data stored in the container. The data has then been stored in a 2D array named data<\/code>.<\/p>\nCreate Weka Instances<\/h2>\n
int numInstances = data[0].length;\nFastVector atts = new FastVector();\nListfor<\/code> loop and use it to iterate over the data items and populate the FastVector data structure with the attributes:<\/p>\nfor(int dim = 0; dim < 2; dim++)\n {\n Attribute current = new Attribute(\"Attribute\" + dim, dim);\n\n if(dim == 0)\n {\n for(int obj = 0; obj < numInstances; obj++)\n {\n instances.add(new SparseInstance(numInstances));\n }\n }\n\n for(int obj = 0; obj < numInstances; obj++)\n {\n instances.get(obj).setValue(current, data[dim][obj]);\n \n }\n atts.addElement(current);\n }<\/code><\/pre>\n<\/div>\nInstance<\/code> class of Weka to generate instances and store them in an Instance variable named newDataset<\/code>.<\/p>\nInstances newDataset = new Instances(\"Dataset\", atts, instances.size());<\/code><\/pre>\n<\/div>\nFit a Linear Regression Model<\/h2>\n
import weka.classifiers.Classifier;\nimport weka.classifiers.Evaluation;\nimport weka.core.Attribute;\nimport weka.core.FastVector;\nimport weka.core.Instance;\nimport weka.core.Instances;\nimport weka.core.SparseInstance;<\/code><\/pre>\n<\/div>\nnewDataset.setClassIndex(1);<\/code><\/pre>\n<\/div>\nLinearRegression()<\/code> function of the Weka library to fit a Linear Regression classifier:<\/p>\nnewDataset.setClassIndex(1);\nfor(Instance inst : instances)\nnewDataset.add(inst);\nClassifier classifier = new weka.classifiers.functions.LinearRegression();\n\nclassifier.buildClassifier(newDataset);\n System.out.println(classifier);<\/code><\/pre>\n<\/div>\nEvaluate and Test Model<\/h2>\n
data2<\/code> and populate it with the test data:<\/p>\n static double[][] data2 = {{14.2,16.4,11.9},{215,325,185}};<\/code><\/pre>\n<\/div>\nint numInstances2 = data2[0].length;\n FastVector atts2 = new FastVector();\n Listfor<\/code> loop to iterate over the data and populate the attributes into the FastVector:<\/p>\nfor(int dim2 = 0; dim2 < 2; dim2++)\n {\n Attribute current2 = new Attribute(\"Attribute\" + dim2, dim2);\n\n if(dim2 == 0)\n {\n for(int obj2 = 0; obj2 < numInstances2; obj2++)\n {\n instances2.add(new SparseInstance(numInstances2));\n }\n }\n\n for(int obj2 = 0; obj2 < numInstances2; obj2++)\n {\n instances2.get(obj2).setValue(current2, data2[dim2][obj2]);\n \n }\n atts2.addElement(current2);\n }<\/code><\/pre>\n<\/div>\nnewDataset2<\/code>:<\/p>\n Instances newDataset2 = new Instances(\"Dataset\", atts2, instances2.size());<\/code><\/pre>\n<\/div>\nnewDataset2.setClassIndex(1);<\/code><\/pre>\n<\/div>\nfor(Instance inst2 : instances2)\n newDataset2.add(inst2);\n\n Evaluation eval = new Evaluation(newDataset);\n eval.evaluateModel(classifier, newDataset2);\n \n System.out.println(eval.toSummaryString());<\/code><\/pre>\n<\/div>\nMake a Prediction<\/h2>\n
Instance pd = newDataset2.lastInstance();\ndouble value = classifier.classifyInstance(pd); \n System.out.println(value);<\/code><\/pre>\n<\/div>\nExecute the Model<\/h2>\n
http:\/\/www.java2s.com\/Code\/Jar\/w\/weka.htm<\/code>
\nI will be using Weka version 3.7.0.<\/p>\ngridstore.jar<\/code> and weka-3-7-0.jar<\/code> files by executing the following commands on the terminal:<\/p>\nexport CLASSPATH=$CLASSPATH:\/usr\/share\/java\/gridstore.jar<\/code><\/pre>\n<\/div>\nexport CLASSPATH=$CLASSPATH:\/mnt\/c\/Users\/user\/Desktop\/weka-3.7.0.jar<\/code><\/pre>\n<\/div>\n.java<\/code> file:<\/p>\njavac IceSales.java<\/code><\/pre>\n<\/div>\njava IceSales<\/code><\/pre>\n<\/div>\n