Customer Relationship Management Data Mining - Essay Example

Relationship Management and Data Mining Task 5 Construction of a ification Matrix Model and the basis of calculation of error rate of 88 records as fraudulent 30 correctly so and 952 as non-fraudulent 920 correctly so. Classification Matrix Predicted Class Actual Class 1 0 1 30 58 0 930 920 Error = (30+920) = 0.49 1938 5.2How moving the cutoff up or down would affect the classification error rate of records that are truly fraudulent and truly non fraudulent Supposing that the overall routine contains cutoff that can be adjusted, the error rate for the records that are either fraudulent or non-fraudulent will vary.

Creating a table as follows will greatly assist in verifying the changescreate the name of the columns one wish to do compute; in this case accuracy and overall error. The left most columns should take the title of Cutoff. In theCorresponding row a formula must be inherited in relevant to the classified matrix cells.Cutoffsthat iwant to evaluate are listed. A range excluding the first column is selected. If i am using Excel software 2007 the following procedure will be used i.e. select data>what if analysis>data Select the cell that changes. Click ok. This table will be automatically complete and will depict the variations.

This shows a lift chart. The lift chart is used to make comparison for thePredictive values given on the base line that depend on average data. The first bar meansthat more performance is noted and gained where specialization and customer focus is achieved. 5.3 Decile wiselifechart for the transaction data model (a). More concentration to customers with divergent needs regards tastes and fashion will make the performance decline gradually. (b).Assuming company A concentrates in selling new clothes whilst company B deals in selling second hand clothing.

In this case company B import second handclothing but has an option of buying disposable items from Company A. Company A on the other hand dispose those clothes that have been late to be sold. Company A will gain higherprofit margins if it focuses its attention to the original customers but will starting losing if itshifts its attention to company B whichhas a wider range of suppliers to choose from. (c).When everything is categorized as non-fraudulent in this case the accuracy will generally increase and the overall error rate will turn to zero.

Illustration: Cutoff   Accuracy   Overall Error     0.49   0.51 1.5         2         2.5         3         2.5         4         4.5         5         5.5         6         6.5         The two models can be widely used in the business circles to gauge Profitability.Flexibility ofthe business. 2.3 Clientele Overall Taste and Preferences In the activity of data mining the most important tool of linear regression is applied. These techniques are used for the purposes of business interference and predictions gauging the performance of the business with the under laying standards.

In the above example where two set of variables are used the profitability of thebusiness venture will be measured by the varying variable which is the customer behavior withthe changing behavior acting against the will and total performance of the venture resulting to decline in the total yields. Multiple regressions are applicable where different and multiple variables can into play. Testing on hypothesis and degree of level of confidence is employed. 2.8. Common based applications in the regression includes measure of sale and products to discounts offered, overall turnover to the price of the product quoted, failure of product usage based onenvironmental conditions etc. 2.10 Customer Focus Most companies have relied on the notion that their products are superior in comparison to the product of the other firms and thus have shifted their attention to their products (product oriented), rather thatcustomers oriented.

The customer margin in this sense drifts away corresponding to the way the firm operates. Understanding the need of the present (existing customers) and newly created ones proofs a must towards excelling in performance. A firm focused at excelling will keep databases holding information regarding customer’s behaviors that will make a firm proactive in making its decisions. A firm will have to employ different software to manage the necessary data. Various techniques will be employed to run the software with each application with different set of problems and challenges.

Companies that have wide marketing techniques have abundant opportunities for survival because their main goal is to retain the high ended customers. 2.11One of the best ways of determining the reliability of goal modeling is by asking queries such as‘is the company determined to attract original customers?’‘Is the firm in need of making her customers profitable?’ ‘Does the company wish to avoid or do away with risky customers?’ ‘Does the firm want to win back lost customers?’ These questions help the management of a company to express her goals in absolute terms. 18.1 Data and Associated forecasting Challenges In this context, series data with an example of autocorrelation brings about profound regression problems.

The approach mostly used in this context is worked out as follows: identifying thecause of the problem discussing the impact of the said problem and eventually coming up withassociated procedure to remedy it. Point and period forecasting are applied as ‘exanti’ and ‘expost’ procedures. Conservative statistical tests are used. Reference Shmueli, G., Patel, N. R., & Bruce, P. C. (2011). Data mining for business intelligence: Concepts, techniques, and applications in Microsoft Office Excel with XLMiner.Wiley.