Data Mining in CRM Framework Management Essay
Data mining can be used to enhance customer relation management. The technique can be used to identify the trends in key business indicators. To do this, the customers will be classified into several clusters using the RFM model in order to identify high value customers. Data mining will be undertaken through association rules algorithms.
Data Mining in CRM Framework for a Health Insurance Company
We will thereafter measure the similarity, differences and modified differences for the mined data and base them into three sets, that is, emerging pattern rule, the unexpected change rule and added rule. We will utilize matching rule thresholds to derive all the types of possible rules and further explore the rules that exhibit significant change based on the degree of change under measurement. Current paper shall utilize data mining techniques to discover the spending patterns and the prevailing trends of behavior change. It will help the management to identify the potential changes in customer preferences. Identifying customer trends will help the company in decision making especially with regard to the policy preferred by the customers or will offer crucial insights with regard to the possible changes to policies, or the introduction of new product or services in line with the customer preferences identified through the data mining process.
Data mining is a computational process aimed at discovering the prevailing patterns from large data set. This approach is very vital in analyzing critical activities for a firm based on the data available from various sources. This project seeks to analyze data mining applications in customer relations management specifically for health insurance company (Rai, 6).
Rationale for Choosing the Topic
Customer relation management is a management technique that utilizes company’s data banks to identify the preferences exhibited by their customers. This enables companies to better offer services by aligning their product offering to the customer preferences. For instance,health insurance company can utilize data mining techniques to identify key trends such as customer’s preferred insurance policy among other trends. Identifying this crucial trend is critical to a company’s survival. Selling insurance policy is the main business for insurance companies. Therefore, identifying the right insurance policies is directly influential to company survival. This, therefore, necessitates that clear customer relation management strategies are designed (Levy 6). Numerous research findings have shown that utilizing company data banks by deducing critical information about customers and identifying the underlying trends using classification algorithms and other data mining techniques is the best technique to manage the existing and or prospecting customers.
This paper seeks to identify key customer trends in health care insurance company. Specifically, the paper will use data mining techniques to identify preferred insurance policy and customer spending patterns. Further, the paper will show how health insurance company can utilize information obtained through data mining techniques to enhance customer relation management. Finally, the paper will show how this process can be used to ensure the company, not just survives but also succeeds.
Customer Relationship Management
Customer relation management is a strategy of customer management that integrates data from sales, marketing and other services. The strategy makes use of procedures and technology to help companies understand their customers better. In addition, the available information can be used to show the customer from different perspectives. Rai (4) identified customer relationship management as a customer centric approach that can help identify customer lifecycle. Rai (34) argued that is the best approach that a company can engage in order to retain customers and achieve customer satisfaction. Customer relation management is a concept that has been used for quite a long time now (Rai, 14).
Evolution of data Mining Techniques in Customer Relation Management
A huge number of businesses in very diverse industries have used this approach albeit, not in the same form as is done today according to Chorianopoulos and Tsiptsis (23). In the past, companies had to identify and keep customer background, spending habits in unconventional ways for instance memorizing by small businesses or keeping hard copy data for large companies. These companies would thereafter choose promotion strategies among other customer related decision based on this information. This is the same principal today. However, the methods of obtaining information and classifying such information have drastically improved. For instance, companies have a lot of information about their customer obtained from various sources which is stored in company’s data banks. This information can further be classified using specialized computer algorithms (Chorianopoulos and Tsiptsis 46).
Using RFM models to Select and Segment Data
RFM (Recency, Frequency and Money Value) is the most used model for selection and segmentation of collected data in order to make analysis easy according to Chorianopoulos and Tsiptsis (35). A firm can use RFM models to sort out the customers targeted from a huge list of customers to enable informed decision making. RFM models of data mining have the ability to make huge influences on a company’s success if the information made available is utilized well by a firm. Sung and Sang used the non-transformed RFM values as direct input variables for building models and then categorized customers into groups using cluster analysis. From the corresponding information marketing strategies and other customer, related activities can be planned based on the different customer clusters identified. On the other hand, Goldman argued that using data mining techniques, for instance using the RFM model of data mining has the effect of avoiding the time wasting and helps companies conserve ‘energy’ by pursuing the customers that the company who would offer the company little or no profits. By doing so, companies can generate better returns by focusing all their resources on the segments of customers who would give the companies better profit margins (Chorianopoulos and Tsiptsis 78).
Role of Information in Fuelling Business Growth
Information plays a very critical in the growth and development of businesses according to research finding by Levy (24). Consequently, the information in businesses databases can be very helpful in ensuring the success of the company. However, this information can only if it is utilized, so trends are deduced from the huge amounts of data. The information in the data bases is obsolete unless elaborate measures are put in place to ensure that it is accessed and grouped such that it shows important indicators and trends in relation to customer experiences and preferences. Once the indicators and trends have been identified, a company can use the information to design strategies that ensure they keep the customers and keep them satisfied too. The success in keeping customers loyal to a brand and satisfied has a direct positive relation with a company’s survival and success regardless of the industry in which the company operates (Levy 23).
Proposed Project and Purpose
The project that I have chosen involves the use of neural networks to predict the customer retention rate and also give the customer insight. The main aim of this because one of the most significant indicators of the marketing campaign is consumer behavior. This project is aimed at studying the behaviors of the customers so as to predict the retention rate. The project is also meant to give the company a model through which it can judge its market position in the future. This calculation gives the company a chance to calculate the profitability margin of all the portfolios that they have invested in. The process involved will be collecting data and then testing the data collected with data mining tools that will be installed in the system of the Java system used.
Neural networks will be the process used to calculate the probability of the customer terminating his or her policy. Once the calculation is done, the results will then be categorized depending on the threshold. The calculation that is most significant result in this process is the ability of the project giving an accurate calculation. For the success of this project to be achieved, the project must give accurate results on the probability. This gives the company an accurate calculation of the performance of the portfolio. The performance of the portfolio then helps the company calculates the profit that is expected (Rai 56).
I choose Neural networks over decision trees mainly because neural networks are more accurate than decision trees. Neural networks enable the users to come up with a classification of the results obtained. This will be very important in Insurance as the clients of the company can be classified depending on their probability of cancelling their policy. Decision trees do not allow the user to classify the data depending on the results obtained. This means that the first intention of creating a classification has failed. It is also very easy to create decision trees from the results obtained from neural networks. This means that neural networks enable the user to create a further analysis if the need arises. Decision trees though easy to use have a weakness when it comes to accuracy that is why I chose neural networks.
The main reason why am proposing this particular project is because it has been successfully applied in many applications. Neural networks have been applied in learning, supervised and unsupervised applications with excellent results. The project also has a time saving quality. Neural networks do not require excessive training to learn the algorithm. Once the algorithm is installed in the Java application being used, it does not require any too much training. The results obtained are also very comprehensible and do not require a lot of time to comprehend them.
It is expected that there may be an objection that may arise if once the project is explained. One objection that may arise will be associated with the cost implementing the project. Neural networks are more expensive to install than decision trees. This means that anyone who looks at the cost may object the project. The response to this will be to explain the limitations that come with decision trees. It will be on no use installing a program that does not meet the needs of the company (Levy 34). The response to this objection will be to explain that the costs of this project will be recoverable easily in the future. All the necessary research has been carried out. I have also considered all the emerging trends that one needs to consider before engaging in any project.
Plan of Activities
There are a number of activities that will be undertaken as illustrated in the work plan below;
Before conducting the data mining process, we will first schedule an interview with a client attached to the insurance company headquarters in Boston. The aim of this interview will be lay ground work and assess the prevailing conditions of the company’s data.
The best way to know how to write good essays is by getting a sample of an essay from competent experts online.
We can give you the essay examples you need for future learning.
Free Essay Examples are here.
After laying the ground work, all the available data- all the data that is available and authorized by the company- will be integrated and explored to identify how useful it can be in identifying or predicting the indicators required.
Building and Validating Models
At this stage, the data mining techniques and models will be built based on the available data and the goals for the project.
Deploying the Models
This task will basically be involved with putting the selected models into practice.
Assessing the Identified Trends
This process will involve assessing the trends identified by the data mining techniques in order to aid in the decision making. As mentioned earlier, the main aim of the project is to identify how data mining can influence customer relationship management. At this stage, the trends outputted will be assessed by the management.
The data to be reviewed will be sourced from the health insurance company data bases. There are numerous sources of data for an insurance company. Any data collected and stored by the company is important and will be used for this project. This include all the data stored in the company’s data bases and any other data collected in other ways, for instance interviews for the purpose of this project (Levy, 34).
Special Tasks and Needs
To establish patterns and trends in a large set of data, in this case from the huge data base of the health insurance company, software with comparatively complex algorithms and data handling techniques are required to be installed in order for them to monitor and highlight the trends. Such software is available in the different operating systems versions.
Data mining is a very delicate operation. Its success relies on the ability to identify the critical variables in the data set and the design of the most effective strategies of utilizing these variables to show or predict trends and patterns. Data mining techniques experts will be required so as to help identify and design the best models for the process.
Data Mining Modeling
This is by far the most important event. Data mining models are the real value behind the whole process. Regardless of the technique used a good data mining models has the ability to provide user-specific solutions. To identify the best model to utilize, the data mining experts sourced earlier will help the health insurance company management to choose the model that best suits their requirements.
Items to be Produced
For the purpose of this project, illustrations covering various operations of the data mining process and test reports will be provided.
This project will be judged based on the applicability and the ability demonstrated to solve the problem identified. The main objective of this project is to demonstrate how data mining techniques and tools can be used by a health care insurance firm to manage its customer relation operations, specifically customer retention rates and offer customer insights. The project success will, therefore, be pegged on the ability to demonstrate ability to use data mining techniques to identify patterns and trends using the most effective models of data mining, the effective application of the model, and the successful use of the trends and patterns identified to influence customer relationship with the insurance company (Rai, 34).
Features to be Assessed
Data Mining Models
The model choice is the most crucial part of the data mining process as identified earlier. The justification and the process of identifying the best model and the subsequent successful use of the model can be assessed as an indicator of the success of the project. The overall effectiveness and success of the project is determined by the success in using the best model.
The outcome of the project will be the production of neural networks with the capability to assess the customer retention rates for health insurance company. The networks will not only have the capability to predict the probability that a customer will continue or terminate a policy, but will also have the capability to indicate the same even with changes to the costs of the premium. Specifically, there will be two neural networks attached to the two clusters of data- the probability contract will be terminated, and the probability contract will not be terminated cluster.
The CRM theory states that the cost of acquiring new customers is always expensive, therefore, the business will always retain their customers by offering them with favorable policies and benefits that are customer friendly. Therefore, the insurance company will use business intelligence theories to ensure minimal cost of operation with increase database of clients. The business intelligence theories help the insurance company to identify new opportunities by developing types of policy covers that ensure customer satisfaction (Provost and Tom, 109).
Predicting Customer Retention Rates
To predict the chances that the company will retain its customers, the neural networks will be programmed to output a number between zero and one. This will indicate the probability that the holder of a health insurance policy will terminate the contract or not. Data will then be categorized into different sets based on a predetermined threshold of 0.5. Any figure above the threshold will be categorized under a threat of termination while any figure below the threshold will be categorized under no threat of termination (Levy, 34).
In this case, the concern is more on the maximization of profitability by selecting the portfolio with the highest returns while at the same time keeping a good market share. To achieve this, the threshold will be set where the predicted rates of termination will be equal to the actual rates of termination. A threshold of 0.204 is set to predict an actual termination rate of 14.7 percent. This will produce an accuracy of 85 %.
Further, to enhance the ability of the neural networks to predict the termination rate when premiums are varied, individual clusters will be subdivided into different bands. On the other hand, to improve the accuracy of the predictions, clusters will be split at the point where accuracy starts to reduce. This will isolate the policy holders based on the significance of the increase in premiums. Two separate neural networks will be deployed for each of the clusters to improve the prediction accuracy.
Data mining techniques can be successfully used to help the company manages its customer relations. The techniques utilize the huge amounts of raw data which is correlated and segmented using specialized computer software into clusters. From this, patterns and trends of consumer behaviors can be deducted. For this project, raw data from health care insurance company was utilized to identify the rates of customer retention as well as to give different insights into consumer behaviors. This information can then be used by management to design customer relation strategies. The successes of these strategies have a direct impact on the success of the business, in this case, the health insurance company.