What Drives Customer Satisfaction in Professional Service Projects?

Student Results:

http://predictive.analyticsight.com/predicting-customer-satisfaction-and-profit-margins/

Introduction.

Professional services are on the rise in the global economy, and due to the private nature of business to business (B2B) transactions, very little is known about the profitability and successfulness of such projects. In partnership with the Technology Services Industry Association (TSIA), we will explore a proprietary database of over 600 IT professional service projects with 50+ metrics. We are tasked with segmenting the projects and uncovering the main drivers behind customer satisfaction. Our work will leverage clustering algorithms, followed by multivariate predictive analysis.

Assignment Details.

Each group has to perform the following set of tasks. All interpretation of results must be embedded into a blog post on this site. You may use this xml diagram for the basis of your work.

Cluster Analysis

task: Interpret the three clusters of projects found in the aforementioned diagram. Instead of copying and pasting tables/graphs from SAS, I encourage you to enter the data into an Excel file and build your own tables. Be sure to cover which variables are included in the analysis, and how each group differs across the variables. When reporting numbers, if it is log variable, report the exponent of that value; if it is a percent variable, report it as X% instead of decimal form.

benefit: TSIA will be able view projects not as one large group, but as three separate groups based on differences across a variety of important project metrics.

Regression Analysis

task: Create 4 regression models for your group’s target variable (either customer satisfaction or margin/margin delta). The first regression will be based on the entire database of projects, while the next three will be completed on each of the aforementioned clusters. Be sure to only include appropriate input variables (to be discussed in class). Lastly, interpret your results, which, for each model, should include at least the following: descriptive statistics of the target (e.g. sample size, average, 25th & 75th percentiles…), estimated relationships between target and significant inputs (either table or graph), and lastly, the accuracy of the model.

benefit: TSIA will be able to understand what factors drive project performance. These factors, and significance of the factors, may change depending on the particular segment. The member companies of this IT association will then be able to fine tune the most important factors so as to increase performance.