Classification and Clustering Analysis using Weka

Classification and Clustering Analysis using Weka

• 1.
  WEKA IT For BusinessIntelligence Ishan Awadhesh 10BM60033 • Term Paper • 19 April 2012 Vinod Gupta School of Management, IIT Kharagpur 1
• 2.
  Table of Contents WEKA! 3 Data Used! 5 Classification Analysis! 6 Cluster Analysis! 11 Other Applications of Weka! 17 References! 17 Vinod Gupta School of Management, IIT Kharagpur 2
• 3.
  WEKA Waikato Environment forKnowledge Analysis DATA MINING TECHNIQUES WEKA is a collection of state-of-the-art machine learning algorithms and data preprocessing tools written in Java, developed at the University of Waikato, New Zealand. It is free software that runs on almost any platform and is available under the GNU General Public License. It has a wide range of applications in various data mining techniques. It provides extensive support for the entire process of experimental data mining, including preparing the input data, evaluating learning schemes statistically, and visualizing the input data and the result of learning. The WEKA workbench includes methods for the main data mining problems: regression, classification, clustering, association rule mining, and attribute selection. It can be used in either of the following two interfaces – •! Command Line Interface (CLI) •! Graphical User Interface (GUI) The WEKA GUI Chooser appears like this – Vinod Gupta School of Management, IIT Kharagpur 3
• 4.
  The buttons canbe used to start the following applications – •Explorer – Environment for exploring data with WEKA. It gives access to all the facilities using menu selection and form filling. •Experimenter – It can be used to get the answer for a question: Which methods and parameter values work best for the given problem? •KnowledgeFlow – Same function as explorer. Supports incremental learning. It allows designing configurations for streamed data processing. Incremental algorithms can be used to process very large datasets. •Simple CLI – It provides a simple Command Line Interface for directly executing WEKA commands. This term paper will demonstrate the following two data mining techniques using WEKA: •Classification •Clustering (Simple K Means) Vinod Gupta School of Management, IIT Kharagpur 4
• 5.
  Data Used The dataused in this paper is Bank Data available in Comma Separated Values format The data contains following fields id - a unique identification number age - age of customer in years (numeric) sex - MALE / FEMALE region - inner_city/rural/suburban/town income- income of customer (numeric) married - is the customer married (YES/NO) children - number of children (numeric) car - does the customer own a car (YES/NO) save_acct - does the customer have a saving account (YES/NO) current_acct - does the customer have a current account (YES/NO) mortgage - does the customer have a mortgage (YES/NO) pep - did customer buy a PEP (Personal Equity Plan) after the last mailing (YES/NO) Vinod Gupta School of Management, IIT Kharagpur 5
• 6.
  Classification Analysis Question "How likelyis person X to buy the new Personal Equity?" By creating a classification tree (a decision tree), the data can be mined to determine the likelihood of this person to buy a new PEP. Possible nodes on the tree would be children, income level, marital status. The attributes of this person can be used against the decision tree to determine the likelihood of him purchasing the Personal Equity Plan. Load the data file Bank_Data.CSV into WEKA. This file contains 900 records of present customers of Bank. We need to divide up our records so some data instances are used to create the model, and some are used to test the model to ensure that we didn't overfit it. Your screen should look like Figure 1 after loading the data. Figure 1.Bank Data Classification in Weka We select the Classify tab, then we select the trees node, then the J48 leaf Vinod Gupta School of Management, IIT Kharagpur 6
• 7.
  Figure 2.Bank DataClassification Algorithm At this point, we are ready to create our model in WEKA. Ensure that Use training set is selected so we use the data set we just loaded to create our model. Click Start and let WEKA run. The output from this model should look like the results in Listing 1. Vinod Gupta School of Management, IIT Kharagpur 7
• 8.
  Listing 1.Output fromWEKA’s classification model What do these numbers mean- Correctly Classified Instances - 92.3333% Incorrectly Classified Instances- 7.6667% False Positives- 29 False Negatives-17 Based on our accuracy rate of 92.3333%, we can say that this is a pretty good model to predict whether a new customer will buy Personal Equity Plan or not. Vinod Gupta School of Management, IIT Kharagpur 8
• 9.
  You can seethe tree by right-clicking on the model you just created, in the result list. On the pop-up menu, select Visualize tree. You'll see the classification tree we just created, although in this example, the visual tree doesn't offer much help. Figure 3. Classification Tree Visualization There's one final step to validating our classification tree, which is to run our test set through the model and ensure that accuracy of the model when evaluating the test set isn't too different from the training set. To do this, in Test options, select the Supplied test set radio button and click Set. Choose the file bmw-test.arff, which contains 1,500 records that were not in the training set we used to create the model. When we click Start this time, WEKA will run this test data set through the model we already created and let us know how the model did. Let's do that, by clicking Start. Below is the output. Vinod Gupta School of Management, IIT Kharagpur 9
• 10.
  Listing 2.Output fromWEKA’s classification model of Test Data Comparing the "Correctly Classified Instances" from this test set (90.5 percent) with the "Correctly Classified Instances" from the training set (92.3333 percent), we see that the accuracy of the model is pretty close, which indicates that the model will not break down with unknown data, or when future data is applied to it. Vinod Gupta School of Management, IIT Kharagpur 10
• 11.
  Cluster Analysis Question: "Whatage groups more likely to buy Personal Equity Plan?" The data can be mined to compare the age of the purchaser of past PEP . From this data, it could be found whether certain age groups (22-30 year olds, for example) have a higher propensity to to go for PEP. The data, when mined, will tend to cluster around certain age groups and certain colors, allowing the user to quickly determine patterns in the data. Load the data file Bank_data.CSV into WEKA using the same steps we used to load data into the Preprocess tab. Take a few minutes to look around the data in this tab. Look at the columns, the attribute data, the distribution of the columns, etc. Your screen should look like Figure 4 after loading the data. Figure 4. Bank cluster data in Weka With this data set, we are looking to create clusters, so instead of clicking on the Classify tab, click on the Cluster tab. Click Choose and select SimpleKMeans from the choices that appear (this will be our preferred method of clustering for this article). Vinod Gupta School of Management, IIT Kharagpur 11
• 12.
  Finally, we wantto adjust the attributes of our cluster algorithm by clicking SimpleKMeans . The only attribute of the algorithm we are interested in adjusting here is the numClusters field, which tells us how many clusters we want to create. Let's change the default value of 2 to 5 for now, but keep these steps in mind later if you want to adjust the number of clusters created. Your WEKA Explorer should look like Figure 5 at this point. Click OK to accept these values. Figure 5. Cluster Attributes At this point, we are ready to run the clustering algorithm. Remember that 100 rows of data with five data clusters would likely take a few hours of computation with a spreadsheet, but WEKA can spit out the answer in less than a second. Your output should look like Listing 3. Vinod Gupta School of Management, IIT Kharagpur 12
• 13.
  Listing 3. ClusterOutput with 5 clusters Vinod Gupta School of Management, IIT Kharagpur 13
• 14.
  Listing 4. ClusterOutput with 10 Clusters Clusters One thing that is clear from the clusters is that behavior of Male are clustered in only 2-3 groups while females behavior are heavily distributed among 7 clusters, so preparing an offering for a specific Description of Clusters- Cluster 0- This group consists of unmarried, mid-income earning females in their early 40’s who live in rural areas. They have on an average two children, no car and personal equity plan but they do have savings and current account. Cluster 1- This group consists of married, high-income earning females in their late 40’s who live in rural areas. They have on an average two children,no car and personal equity plan but they do have savings and current account. Cluster 2- This group consists of married, low-income earning females in their early 40’s who live in inner city. They have on an average one child, no car and savings account but they do have current account and personal equity plan. Vinod Gupta School of Management, IIT Kharagpur 14
• 15.
  Cluster 3- Thisgroup consists of married, low-income earning females in their early 30’s who live in town. They have on an average one or two children, no car, savings account and personal equity plan but they do have current account. Cluster 4- This group consists of married, mid-income earning males in their late 30’s who live in inner city. They have on an average one or no child, no savings account but they do have personal equity plan, savings & current account. Cluster 5- This group consists of unmarried, high-income earning males in their early 40’s who live in town. They have on an average one or no child, they have car, personal equity plan, savings & current account. Cluster 6- This group consists of married, mid-income earning females in their early 40’s who live in inner city. They mostly don’t have ant child, they do not have any savings account and personal equity plan but they do have current account. Cluster 7- This group consists of unmarried, high-income earning females in their mid 40’s who live in inner city. They have on an average one or two child, no car and personal equity plan but they do have savings & current account. Cluster 8- This group consists of unmarried, high-income earning females in their mid 40’s who live in town. They have on an average one or no child, no personal equity plan but they do have car, savings & current account. Cluster 9- This group consists of married, mid-income earning males in their early 40’s who live in inner city. They have on an average one or two children, no car, personal equity plan and current account but they do have savings account. Vinod Gupta School of Management, IIT Kharagpur 15
• 16.
  One other interestingway to examine the data in these clusters is to inspect it visually. To do this, you should right-click on theResult List section of the Cluster tab . One of the options from this pop-up menu is Visualize Cluster Assignments. A window will pop up that lets you play with the results and see them visually. For this example, change the X axis to be income (Num), the Y axis to children (Num), and the Color to Cluster (Nom). This will show us in a chart how the clusters are grouped in terms of income and no’ of children. Also, turn up the "Jitter" to about three-fourths of the way maxed out, which will artificially scatter the plot points to allow us to see them more easily. Figure 6. Cluster Visual Inspection Vinod Gupta School of Management, IIT Kharagpur 16
• 17.
  Other Applications ofWeka •DISCRETIZATION •REGRESSION •NEAREST NEIGHBOR References https://www.ibm.com/developerworks/opensource/library/os-weka2/ http://maya.cs.depaul.edu/classes/ect584/weka/preprocess.html http://www.cs.waikato.ac.nz/~ml/weka/ Vinod Gupta School of Management, IIT Kharagpur 17