IS428 AY2019-20T2 Assign ARINO ANG JIA LER

Overview

Background

Every two years, SMU Libraries conduct a comprehensive survey in which faculty, students and staff have the opportunity to rate various aspects of SMU library's services. The objective of the survey is to provide these libraries with input to help improve existing services as well as information regarding anticipating emerging needs of SMU faculty, students (undergraduates and postgraduates), and staff. The latest survey ended on 17th February 2020. The past reports are mainly made-up of large numbers of pages of tables, which are difficult to comprehend. The difficulty in comprehension has resulted in the inability to draw useful information that could help SMU Libraries in making the necessary improvements or changes.

Objective

The aim of this project is to design a better dashboard/report that would allow SMU Libraries to extract valuable insights from the perspective of undergraduate students, postgraduate students, faculty, and staff. The dashboard designs would mainly surround the responses of each assessment item in the survey (e.g. services provided by the library) as well as the net promoter score. There would also be some focus on the sub-attributes of people who provide certain responses (e.g. attributes of respondents who are unhappy with a certain service provided by the library).

Data preparation

Dataset

The dataset provided in this project is the 2018 Library Survey dataset. In this dataset, there are 89 variables and 2639 responses. 26 of the 88 variables are not applicable (NA01 – NA26), thus, we are excluding these 26 variables. There is 1 identifier variable used to uniquely identify each response (ResponseID)

For the remaining 62 variables, 26 of them measures the importance of each assessment (i01 – i26), and the other 26, based on the same questions, measures the performance of each assessment (P01 – P26). There is 1 variable on overall satisfaction of the library, measured using performance as well (P27). All assessment variables are based on Likert Scale data.

The remaining 9 variables provides information on respondent behaviour (Campus, HowOftenL, HowOftenC, HowOftenW), characteristics (StudyArea, Position, ID), net promoter score (NPS1) and an open-ended comment section (Comment1).

Data cleaning & transformation

Screenshots	Description
Figure 1.0: Hiding of columns	Na01 to Na26 columns are hidden and not used in the analysis and the design of dashboard.
Figure 2.0: Pivoting of Columns	Pivot all the assessment items, i01 – i26, and P01 – P27.
Figure 3.0: Renamed Columns Figure 3.1: Aliases for Assessment Items	Pivot Field Name and Pivot Field Value are renamed into Library Assessment Items and Assessment Score respectively. Each value of Library Assessment Items is also given an alias that best represents the question in the survey. Aliases are labelled (I) at the back to represent Importance and (P) to represent performance. This step is necessary to identify whether an Assessment Score represents Importance or Performance.
Figure 4.0: Columns with new Aliases Figure 4.1: Aliases for Campus Figure 4.2: Aliases for Position Figure 4.3: Aliases for Study Area	The values of Campus, Position, and Study Area are also given aliases for better comprehension.
Figure 5.0: Renamed Columns Figure 5.1: Aliases for 3 values of frequency	How Often L, How Often C, How Often W, and Nps1 are renamed to Library visit frequency, Campus visit frequency, Library resource access frequency, and Net Promoter Score respectively. The 3 values of frequency are also given aliases for better comprehension. All 3 frequency variables have values with the same properties, visit frequency decreases as value is increased (e.g. 1 – daily, 5 – never).
Figure 6.0: Groups for Library Assessment Items Figure 6.1: Groups for Net Promoter Score Figure 6.2: Groups for Position	Groups are also created for Library Assessment Items, Net Promoter Score, and Position. Some of the groups would function as filters, others for calculation and hierarchical purposes.
Figure 7.0: Collection of Sets Figure 7.1: Sample of a set using Postgraduate Figure 7.2: Sample of filter interface using Postgraduate Set	Sets are created to act as filters. For example, users who are using the Postgraduate dashboard should only be able to select either Masters or Doctoral from Position and not Year 1 as Year 1s are not Postgraduates. For the remaining Undergraduate, Faculty, and Staff Sets, the values that should not be part of the set are excluded. The Assessment Set was created initially to exclude the Overall Satisfaction values. However, throughout the creation of the dashboard, I decided to integrate Overall Satisfaction into the filters. This step is not necessary.
Figure 8.0: The 3 new measures created for NPS Figure 8.1: Formula for Detractors Figure 8.2: Formula for Promoters Figure 8.3: Formula for NPS	3 new measures are created for calculation of Net Promoter Score as I want to display the Net Promoter Score value alongside its distribution.
Figure 9.0: Sheet used for creation and verification of Diverging stacked bar chart variables Figure 9.1: Collection of Measures created for the Diverging stacked bar chart	Following instructions provided during class on creating diverging stacked bar charts by Prof Kam, I have dedicated a worksheet for the creation and verification of variables required for the chart. The 6 variables required are Count Negative, Gantt Percent, Gantt Start, Percentage, Total Count, and Total Count Negative.
Figure 10.0: Collection of Sorting variables Figure 10.1: Formula for Score Sort (Highest) Figure 10.2: Formula for Score Sort (Lowest)	2 variables Score Sort (Highest) and Score Sort (Lowest) are created to sort the diverging stacked bar charts later on. Score Sort (Highest) allows the stacked bar chart to be sorted by the highest number of positive responses to the lowest number of positive responses (e.g. For the assessment items pertaining to importance, I want to display the most important assessment items at the top, and the least important assessment items at the bottom) Score Sort (Lowest) allows the stacked bar chart to be sorted by the highest number of negative responses to the lowest number of negative responses (e.g. For the assessment items pertaining to performance, I want to display the lowest performing assessment items at the top, and the highest performing assessment items at the bottom) These 2 variables would help to answer questions like “What are the Top 5 most important components of the library?” and “What are the Top 5 most underperforming components of the library?” Higher weightage is given to Assessment Scores at extreme ends (e.g. 1 and 7). This helps to magnify Assessment Items that users have stronger feelings towards (e.g. importance score of 7 is assumed to invoke feelings of higher importance than an importance score of 5)
Figure 11.0: Global filter for the entire Dashboard	As there are several Null values for Assessment Scores that is not useful for the analysis, a global filter is created to filter out all Assessment Scores with Null values.

Dashboard

Storyboard Link

Link to storyboard: https://public.tableau.com/profile/arino#!/vizhome/IS428_Assignment_Tableau_Arino/Storyboard

Storyboard Overview

Analysis & Insights

For the following sections, we would explore the data based on each major group of library users. However, there would be some cross referencing and comparison across the major groups. There would also be different levels of analysis for each major group or subgroup depending on what is relevant.