Use data mining tools (Weka) to enhance a marketing exercise - Assignment Example

? DATA MINING TOOLS al Affiliation) Key words: Data mining, Model Table of Contents Introduction 3 2.Data pre-processing 4 3. Costs 13 5. Recommendation 20 6. Conclusion. 21 Reference 22 1. Introduction The paper discusses data mining tool used by Tayko, an international company selling electronic devices and components. As a data analytics, I have designed a model to help the company target their promotional mail out. Tayko Company has a series of names in their mailing list by capturing their logging and their internet enquiries targeting the electronic market. Tayko company is almost dispatching their next mail out, the company want to flag out clients who are promotional receptive. This is important because it will help them reduce their cost and also maintain a good relationship with their customers by not becoming a bother to those who do not need the promotion. The company wishes to come up with a better and a targeted list by selecting their clients randomly and sending the trial promotion to them. The paper therefore, mine the data to assist Tayko identify the attributes of companies that show that they will respond positively to the mail-out. This will help improve the performance of mailing promotion. The attitude of the customer toward the product, and the perception of the company of origin are crucial factors when assessing the receptiveness of the customer. As the figures of younger, and high income clients with increasing demand for electronic products continues to grow, the market become interesting and the need on assessing the receptiveness of the clients to the company’s brands and business becomes very crucial. The receptive of the customers to the company’s goods is not considered a conventional framework when selecting the market. However it has crucial implications for the marketing department in sensitizing and assessing the standard, easily accessible risk indicators of commercials used. Tayko introduces the customer receptiveness as an added criterion in the specialized mechanism to the assessment of its customer’s relation. 2. Data pre-processing a) Therefore, the best indicator of the client’s receptiveness to the promotion is attribute number 25 describing the amount spent by client in test mailing in terms of dollars. b) Another attribute that could be selected as this, is the attribute that highlighted the client made purchases in test. The “test purchase request” will document the service pursuant of sales and has information on the clients name, address, corporation, and firm where the customer’s request. Additionally, it contains the office mailing address, signature, title, name, and the telephone numbers for the client making the request. The attribute is also important because it contains the statement of the conduct nature under investigation. Also, it is crucial because it contains the statement that the service must be tendered at the place and time. c) There are other attributes that are not important in this project. Such attributes includes, “How many days ago was first update to customer record”, and the sequence number for the customer (Han & Kambe, 2006). The meaningless attributes creates a valued difference between the brands, and during the process. When the meaningless attributes are added they change the decision consumer structure, majorly if the differentiated attribute is hard to evaluate. This may make the consumers to infer the value of attributes. d) Some of the classification model that I designed using the Weka classifier is as shown in the figure below. The above is the visualization of the mailing promotion model. Data @attribute seq numeric @attribute US binary @attribute Freq numeric @attribute web_order {1, 0} @attribute Gender=male numeric @attribute Address_is_res binary @attribute Purchase {1, 0} @data 4, 200210, 200601, 0 5, 200301, 200601, 1 Calculation === Run information === Scheme: weka.classifiers.rules.ZeroR Relation: Ass3Data67_33 Instances: 1501 Attributes: 25 Seq US Source_1 Source_2 Source_3 Source_4 Source_5 Source_6 Source_7 Source_8 Source_9 Source_10 Source_11 Source_12 Source_13 Source_14 Source_15 Freq Last update First update Web Male res_addr purch spend Test mode: 10-fold cross-validation === Classifier model (full training set) === ZeroR predicts class value: 0 Time taken to build model: 0 seconds === Stratified cross-validation === === Summary === Correctly Classified Instances 1000 66.6223 % Incorrectly Classified Instances 501 33.3777 % Kappa statistic 0 Mean absolute error 0.4448 Root mean squared error 0.4716 Relative absolute error 100 % Root relative squared error 100 % Total Number of Instances 1501 === Detailed Accuracy By Class === TP Rate FP Rate Precision Recall F-Measure ROC Area Class 1 1 0.666 1 0.8 0.499 0 0 0 0 0 0 0.499 1 Weighted Avg. 0.666 0.666 0.444 0.666 0.533 0.499 === Confusion Matrix === a b Read More