Data Analysis: Predicting Customer Satisfaction And Likelihood Of Building Strategic Alliances

Dependent Variables

Business operations are controlled by different factors; among which include technology and changes in demand. Auspaper is a paper company that has been operating comfortably by selling printing paper to various countries across the globe. Technological improvements have changed lifestyles of individuals in the society. Some of the changes that have been embraced include the mode of reading magazines and newspapers. More people are reading the dailies among other reading materials over the technological gadgets such as smartphones, tablets and computers. The improvements have changed the demand and supply levels for Auspaper Company, hence reducing their market achievements. As a result of these changes, Auspaper has focused on developing strategic alliances with magazines and newspaper printing companies to maintain their sales. This paper includes a report on data analysis that focuses on the satisfaction of a customer, hence evaluating their willingness to engage in a strategic alliance.

As stated in the introduction section, this paper will focus on developing a model that can predict the satisfaction level of customers and the probability of engaging into a strategic business alliance. Therefore, multiple linear and logistic regression statistical tools have been used to answering these questions. The satisfaction variable is continuous, which meets a requirement for multiple regression. The average satisfaction level for the customers was 6.952, which is above average because the score runs from 0 to 10. Auspaper services and product quality seems to be above average because there were not customers whose view were below 4, which can be termed as under-satisfaction. The minimum satisfaction score for the customers was 4.7 and the maximum was 9.9.

The variable representing the views of the customers based on their consideration to engage in a strategic alliance is a dummy variable. 57% of the customers stated that they did not want to build a strategic business alliance against 43% who would accept the agreement. The high number of customers rejecting the offer is because of the reducing demand of published reading papers and magazines. The strategic alliance variable will be used on a logistic variable to determine the probability of a customer accepting the agreement based on different score levels for the significant variables.  

The data set included a set of continuous and categorical variables that can be used in developing a prediction model for customer satisfaction level for the customers. The first requirement for the continuous variables is they should be significantly correlating with the response variable. Therefore, a covariance matrix was developed and it was found that five (5) variables had at least a correlation coefficient of 0.5 with satisfaction variable. These variables include product quality, complaint resolution, product line, order & billing and delivery speed. Also, the sales force image could be categorised as a variable that has a moderate correlation with satisfaction level because it coefficient was 0.4779. Therefore, six of the continuous variables could be used in developing the model for predicting customer satisfaction level. The categorical variables were also analysed and it was observed that customer type was approximately equally distributed among the three groups. The industry types were either magazines or newspaper, which were distributed in the data. Also of the firm was approximately represented in the data. The customer location was either Outside ANZ or within ANZ. More customers from outside ANZ were represented in the sample than those from ANZ.  Therefore, the categorical variables would all be included in the model, hence determining their significance based on their p-values.

Identifying customer satisfaction predictors

This part will make used of the identified variables in the section above to build a predictive model for prediction of customer satisfaction level. These variables include; product quality, complaint resolution, product line, the image of product, billing, and speed of delivery, type of customer, industry type, size region and system of distribution. The model was developed and it was found to be generally significant with a p-value less than 0.0001. In addition, the model achieved an R-squared value of 0.8496, which shows the model was well fitted and the set of variables highly contribute to the changes in the response variable. After the categorical variable that had more than two levels were split, a total of 12 predictor variables were achieved, creating a significant set to be used in predicting the satisfaction. Among all the 12 predictor variable, 5 were found to be statistically insignificant at 95% confidence level. Moreover, because the model was highly significant, I decided to leave it unchanged and use all these variables in the prediction. The model can be displayed as shown below.

The above equation could be used in predicting the satisfaction level of a customer and the set of predictors explains 84.96% of the variations experienced on the response (satisfaction) variable.

The significance of interaction effect can be measured based on the changes observed as a result of its inclusion. The interaction between product line and location of a customer improved the model from an R-squared value of 45.22% to 50.29%. This indicates that the interaction effect moderates the model’s significance. Both of the predictor variables and the interaction are statistically significant at 95% confidence level.

Product line and image of the Salesforce are thought to be significant variables that can be used is determining the likelihood of customer building the strategic alliance. The coefficients of the logistic model included 8.988 for the intercept, -0.816 for the product line and -0.735 for the image variable. Therefore, the probability of a customer building a strategic alliance with Auspaper would obtain using the formula below:-

Using the formula above, the probability of achieving strategic alliance for neutral opinions on the product line and Salesforce image would be as shown below:-

The logistic model’s sensitivity is 79.82% and specificity of 63.95%. Therefore, I can conclude that the study was more sensitive; whereby more true positives were obtained that the true negatives. The model’s coefficients were 12.587 for the intercept, -1.055 and -0.866 for product quality and price flexibility respectively. These coefficients were used in determining the probabilities for having a strategic alliance for different product quality and price flexibility opinion score levels. It was observed that price flexibility score of 0 a price quality of 1 gathered a zero probability of having a strategic alliance. However, a probability close to one was obtained for a price flexibility and product quality of 10.

A simple forecasting model was used to establish a time series model to predict the turnover for the 2^nd, 3^rd and 4^th quarters in 2017. The equation below was obtained after fitting a linear forecast model on the data and it will be used in the prediction process.

Order an Essay Now & Get These Features For Free:

Turnitin Report

Formatting

Title Page

Citation

Outline

Place an Order