What Are the Main Kinds of Attrition - Assignment Example

Predictive Analytic Student’s Name Institutional Affiliation Predictive Analytic Part A What are the main kinds of attrition? What is the differences and significance of these? Three types of attrition exist between the customers and the company. Involuntary attrition occurs when the firm terminates the bond that links it to the customers. In this case, it arises when the consumer is not able to pay the bills making the company to suffer a loss. Expected attrition occurs when the client does not exist in the market as expected by the company. The sales volume thus reduces due to the absence of the customer. For example, when a person develops, he will tend to change the eating habits in terms of changing the foodstuffs. As a result, they will not need the bay food any more. The firm, therefore, will seek other types of products or changes their mode of production to keep the demand of the customer intact (Berry & Linoff, 2010). Forced attrition involves when the company decided to employ certain measures to retain the customer. For instance, a company may lower the price of the product to lure or retain the clients of a particular commodity. Some of the clients may run away from buying a product because of the high prices. In this instance, the consumer moves out of the shopping complex involuntarily. The person may be willing to purchase the product but encounters financial constraints. Evidently, many people tend to confuse these three types of attrition. It is important to note the differences to know the future performance of the company as well as determination of the production prospects. Through these attritions, one can adjust their way of performance to avoid losing their potential customers. The worst point of confusing these types of attritions is that the company may use money to retain the customer but later become a bad debtor (Berry & Linoff, 2010). The firm loses twice because of the cost of retaining the customer and that of providing for the bad debt. Therefore, data mining asserts that it is important to address both the voluntary and involuntary attritions due the changing behaviors of the customers. The two attritions will cover both the adverse and positive sides of the company. That is, all the clients are at risk for both types of attritions at varying degrees. There are two main kinds of attrition models – modeling attrition as a binary outcome, estimating customer lifetime: Types of Attrition Models Predicting who Will Leave In this type of attrition model, it is important to note the period for the binary outcome. Usually, the questions of who will leave the company in future or expected dates are hard to answer because nearly everybody would want to move out of the company. In this regard, binary outcome attrition models have definite timeframe say 60 to 90 days or even one year. Clearly, the horizon cannot be that short because there will be chance to act on the predictions of the models. The determination of the customers to stay or leave the market arises from the prediction mixtures like probabilities, decision trees and neural networks. In most of the cases, the companies use flag to show the population of the customer that could leave the market after a certain period. As a result, the modeling task is to evaluate the number of the buyers that remains in the market to that number which persists in purchasing the products of the firm. People like this type of attrition model because it ranks the buyers by their likelihood of leaving the market. The results from such evaluation determine the retention of the consumers in the market. The individuals with voluntary attrition who attain the points that surpass the threshold can form part of the retention program. On the other hand, the individuals with involuntary attrition scores that exceed the threshold can be under the watch list. Expertise predicts the behavior of the customers using the items known about the customer at the acquisition time (Berry & Linoff, 2010). Further, the predictors also evaluate certain elements that existed during the relationship with the customer like the challenges of late payment, changing bills customer demographics. This type of modeling attrition avails the information on how to reduce future attrition by developing some techniques such as lesser attrition-prone customers. Preferably, the second-class attrition also provides the insight on how to mitigate the risks for the clients that exist in the market. Predicting how long the Customers Will Stay in the Market The basic idea in the type of attrition model is the calculation of the input variables that help him stay in the market and determination of the factors that may make him leave the market in future. The calculation encompasses the variables like geography, the social class as well as acquisition channel. Most of the researchers refer to this probability as hazard probability because it determines the unfavorable factors that may influence the purchasing power of the consumer. The intervening hazards are used to calculate the possible chances that can make the customer reach the anticipated date in the market. Survival analysis is a term used to using analysis and analytics techniques to understand ‘when to start worrying’ about customers (in a business/marketing context). Compare and contrast data volumes used for survival analysis in business applications with medical and manufacturing applications. Survival analysis is vital in understanding of the customers in the market. It first originated from the medical department and later used by the manufacturing industries. The use of this technique helps in the realization of the prospect of the industry depending on the behaviors of the customers. Precisely, this mechanism determines the right time to start worrying about the customers in the contemporary market. As such, it displays the factors that accelerate certain behaviors of the consumers (Berry & Linoff, 2010). To illustrate this, the expertise uses the hazard and survival curves to shows the purchasing patterns of the customers in the market. The curves also addresses the concern of the quantity of the company should produce to match the demand of the clients. Likewise, the technique also predicts when the customers are likely to leave the market. Some of the patterns showed by the curves include the frequency of purchasing a given product to determine the time a customer will take to purchase the product again. When the customer fails to turn over for the products as per the predictions of the management, then the customer is likely to avoid that commodity. Survival analysis as a proponent is a vital facet in the determination of the customers’ behavior. It describes the clients as the source of the business information. They make the management to change their ways of doing things and set relevant strategies to achieve the goals of the business. The mystery that the customer leaves behind after walking out of the market is the driving force in the determination of the targets in the firm. As a result, all the business pundits respect the value of the customer’s loyalty. The assumption made by the company during the survival analysis is that the longer the customer takes in the market the lower the risk of stop buying the products of the company (Berry & Linoff, 2010). However, the assumption may not underscore because of the various hazards involved in the market. The level of competition among other factors may make the customer to change his mind. The survival of the customer is the best way to measure the retention of the customers as far as the company is concerned with attracting the attention of the customers. What Survival Curves Reveal Survival curve as a quantitative technique shows the extent at which the customers are to survive up to a particular time in the market depending on the historical data on the duration of survival in the past situations. Usually, the curve begins from the 100% and descends to various points on the scale depending on the behavior of the customer. It can extend to zero but that does not imply that the buyers with the longest tenures remain active in the market at all time. Moreover, this technique reflects the survival value of the customer. To compare the customers’ survival is by determining the half number of the customers that leaves the market; the half-life of the clients. This is vital because few customers with the longest or the shortest lifetime does not affect it. The Calculation and Relevance of Average Tenure from a Survival Curve The expertise use the customer’s half-life for comparisons because it easier to compute or calculate. However, there are various weaknesses of this techniques; it does not address the concern of the average and the worth of the customer during the computation. Answering those questions requires the analysis of average worth of the customer depending on the changes of time per client as well as the average survival for all the buyers (Berry & Linoff, 2010). According to the research, the median cannot provide the data related to the parameter because it only describes one activity of the customer at the middle stage. The person who leaves after the other half of the previous customers left the market. The area under the retention curve represents the remaining halftime. The expertise use calculus to find the population of the customer that is out of the market and that which remains. Notably, the average in this case represents that of the entire period under the survival curve. Part B Seven Reasons Why We Need Predictive Analytic The emergence of advanced technology in the contemporary market has led to high level of competition. For this reason, all the companies should strive to improve both the quality and quantity of their products to win the interest of the customers. Predictive analytics help in the determination of the product that would compete with other producers as afar as the level of competition is concerned in the market. It aids in the value proposition that makes the managers to make the right decision concerning the product to compete in the market. it also reveals the weaknesses as well as the strengths of the other competitors hence the business is able to make the necessary changes to win the market. Increasing the Sales Volume and Retaining of the Customers Predictive analytics helps in the application of marketing and sales to generate profits in the firm. Outstandingly, the aim of every business entity is to increase the sales volume and maintain the customers in the market. It underscores the behaviors of the customers in relation to purchasing power and the consumption level (Berry & Linoff, 2010). The company then responds by producing the commodities that meet the demand of the customers in the market. Maintenance of Integrity and Fraud in the Business The current management connotations encourage the misuse of the resources as well as fraud in the most of the industries. For this reason, there is a need to negate the vice by employing proper measures to combat such activities within the industry verticals. Predictive analytics helps in the creation of the model that ranks the records of the organization. This movement assists in the detection of the malpractices in the organization. Apparently, whenever there is alteration of documents during transactions, it raises alarm and the unit control detects analyses the changes. Some of these activities that predictive analytics has played vital role on include over quoting of the prices, delivering more items among other activities. Advancement of Business Capacity One of the most frontiers of predictive analytic is improving the efficiency of the business activities. Preferably, it advocates for the changes necessary for positive improvement. The industry aims to deliver efficiently with the help of predictive analytics. Matching the Demand of the Consumers Predictive analytics is a tool that focuses on benefiting the consumers. It compromises the types of the goods and the expectation of the customers. It ensures that the company produces relevant goods and services with sufficient qualities and quantities as the customer’s demand. Employment of the Advanced Analytics Predictive analytics determine the technological changes in the company. Ti relates the past activities with the current environmental changes and suggests the best technology the industry should employ to match the level of competition. The technology should be able to produce goods of the highest quality. Determination of the Business Intelligence Clearly, implementation of the decisions in the business determines the progress of the company. Predictive analytics avail recommends the use of the analytic model in ensuring proper implementation of policies. It therefore provides the strategies and generates an insight of the company. Predictive Analytics White Paper – PA for Insurers’ Insurance organizations alongside other institutions are crucial predictors of the future progress of the organizations. Most of the industries operate under the environment of uncertainty making them to embrace predictions on various environmental changes (Fitz-enz & Mattox, 2014). They forecast the different prospects and evaluate the premiums in terms of age, occupation history and the level of income of a person. Predictive analytics refers to the application of relevant models and techniques in predicting the future events. Drivers of Insurer’s Use of Predictive Analytics Various drivers predict the future events that the insurance industries use. Some of them include technological advances, data availability, the desire of insures’ growth in the contemporary market and the urge of insures’ to compete with other companies. These drivers are crucial in the determination of the future events (Berry & Linoff, 2010). For instance, insurance companies use computers in the calculation of premiums and entry of data depending on the quantity of data. Moreover, the companies save their money in form of premiums, which help them later to expand their activities. The use of Predictive Analytics in Insurance Companies The insurance companies find it easy to scan the market using predictive analytics. Usually, it helps in showing the policy purchasing patterns in the market. Further, predictive analytic help insurance company to filter out the customers that do not meet the threshold of the policies. As a result, there is no much time wastage by both the employee and the company resources. Finally, the insurance companies use this technique to detect potentially fraudulent claims. Insurance and the Seven Reasons for Having Predictive Analytics Just like predictive analytics important in the economy, insurance is crucial in shaping the activities of the companies. For instance, saving the money inform of premium enables the company’s expansion of activities because of the additional capital. Insurance also reduces the risks of mismanagement of funds because it takes control of the savings. Insurance institutions offers the information required in the implementation of the future planning (Cappelli, Newmark & National Bureau of Economic Research, 2001). In addition, the companies may take covers for their customers to maintain the number of the buyers. Finally, insurance companies has the consultants that take charge in ensuring that the companies avoid future risks by operating certain projects. Part C 1. It is not advisable to perform the imputation after the partitioning since Enterprise Miner utilizes the imputation data from the training data and confirms them to the validation data. This mechanism helps in ensuring that the two sets of data are comparable. For example, the missing variable namely TOT_REV_AMT and TOT_PROF_AMT contained high rate of missing values due to lack of collection of data for the customers (Fitz-enz & Mattox, 2014). The collection of data follows the determination of the customers to churn in the past. The missing customers have high probability of churning. Question 2 After preparing the variable for data selection, there is a need to eliminate redundant variables using the variable clustering. To achieve this target, one should add a variable clustering node to the diagram to change the faulty components. After that, one should click on the explore button and drag the variable clustering node into the diagram (Cappelli, Newmark & National Bureau of Economic Research, 2001). At this stage, a person should conclude by connecting replacement node to the variable clustering node. In addition, the variable clustering node also enables the running the program to achieve the desired results. It controls the Yes button to run the system. Question 3 Usually, when operating the logistic regression, people encounter a problem so-called complete separation or quasi-completed separation. A complete separation occurs when the final variable separates a predictor variable or the mixture of predictor variables. Y X1 X2 0 1 3 0 2 2 0 3 -1 0 3 -1 1 5 2 1 6 4 1 10 1 1 11 0 In the above example, Y is the outcome variable, X1 and X2 predictable variables. Observable, when Y=0, all the values of X are equal or less than 3 while when Y=1, all the values of X are 1 or greater than 3. Outstandingly, complete separation occurs arises due to various reasons. For instance, when using categorical variables coded with indicators, complete separation can arise (Boire, 2013). When one tries to fit a logistic regression model of Y on X1 and X2, there is no maximum likelihood estimate for X1. The explanation behind the notion is that the larger the coefficient, the bigger the likelihood. Illustration using the data from SAS Data t; Input Y X1 X2; Cards; 0 1 3 0 2 2 0 3 -1 0 3 -1 1 5 2 1 6 4 1 10 1 1 11 0 ; run; proc logistic data = t descending; model y = x1 x2; run; (some output omitted) The programming channel Complete separation of data points detected. “WARNING: The maximum likelihood estimate does not exist” “WARNING: The LOGISTIC procedure continues in spite of the above warning. Results shown are based on the last maximum likelihood iteration”. “Validity of the model fit is questionable” “Model Fit Statistics” Intercept Intercept and Criterion Only Covariates AIC 13.090 6.005 SC 13.170 6.244 -2 Log L 11.090 0.005 “WARNING: The validity of the model fit is questionable”. “Testing Global Null Hypothesis: BETA=0” Test Chi-Square DF Pr > ChiSq Likelihood Ratio 11.0850 2 0.0039 Score 6.8932 2 0.0319 Wald 0.1302 2 0.9370 “Analysis of Maximum Likelihood Estimates” Standard Wald Parameter DF Estimate Error Chi-Square Pr > ChiSq Intercept 1 -20.7083 73.7757 0.0788 0.7789 X1 1 4.4921 12.7425 0.1243 0.7244 X2 1 2.3960 27.9875 0.0073 0.9318 Question 4 When they are the same, there are certain components that undergo scanning to determine some basic differences. For instance, there is detection of their origins to give the proper analysis. It is most likely that the time determines the differences in their originality (Zhang & Zhang, 2015). Moreover, there is indication of time on the scale in terms of months and days. References Bari, A., Chaouchi, M., & Jung, T. (2014). Predictive analytics for dummies. Hoboken, NJ: John Wiley & Sons. Berry, M. J., & Linoff, G. (2010). Data mining techniques: For marketing, sales, and customer relationship management. Indianapolis: Wiley. Boire, R. (2013). Predictive analytics: The power to predict who will click, buy, lie, or die. J Market Anal, 1(3), 184-185. doi:10.1057/jma.2013.14 Cappelli, P., Newmark, D., & National Bureau of Economic Research. (2001). External job churning and internal job flexibility. Cambridge, MA: National Bureau of Economic Research. Fitz-enz, J., & Mattox, J. (2014). Predictive analytics for human resources. Zhang, Z., & Zhang, P. (2015). Seeing around the corner: an analytic approach for predictive maintenance using sensor data. Journal of Management Analytics, 2(4), 333-350. doi:10.1080/23270012.2015.1086704 Read More