Imperial University ACTEM Events

Data and Methods

The dataset was extracted from an online questionnaire, the event host used Google Sheets to input the students' answers into Excel. Once the data was handed to me, the next step was to transform the data. The cleaning of data included removing null, from multiple rows, correcting text format and deleting unnecessary rows which were not relevant to the hypothesis of the data. In addition, I had to merge two Excel worksheets into one sheet. I received 2 CSV files, both files were different as one of the files had qualitative data, and the other file had quantitative data. So I had to merge the data from both CSV files into one CSV file to represent both quantitative and qualitative data. After the data cleaning was completed in Excel, the data was then loaded to Power query to do further data cleaning and filtering. To begin I used the data preview to analyse column quality, column distribution and column profile.

Once this step was completed, I also checked the right format was represented for data in each column. For example, checking text fields has text data and number fields has number data. After the data cleaning was completed it was time to shape the data.

To begin filtering the data, I copied the query from the original dataset and named its responses. To filter the data further a naming convention":–TEXT" was input to the Filed header that had long text data. The reasoning for this was to use this as a filter so that long text data does not show on the matrix as it will make the matrix look messy.

Unpivot other column command was used on the ID Filed, this allowed for all the questions to be split into columns, which were named questions. Also, the other column that was unpivot was named responses. The spilt column on the delimiter command was used on the question column to further split the column. Then the new column which was split on delimited was then called question type. Now the question type column can be used as a filter in the report.

A matrix was the best way to visualise the data. As a matrix visual icon automatically aggregates the data and allows the user to drill down to see more information about the data.

After that, I create a measure for the count and percentage please see, the Dax formula I used to do this below.

1 Count Responses = COUNTROWS(Responses)

% Of Total =

var count response = [Count Responses]

var response = CALCULATE([Count Responses], ALL(Responses),Responses[Question]= SELECTEDVALUE(Responses[Question]))

RETURN Divide(countresponse,allresponse)

Results and key findings

From looking at the results of the survey data, it is safe to say that the event was a success against the key metric of the event rating.

From looking at a score of Delivery of session rating, learning environment rating, engagement of session rating and content quality rating. The event scored over 4.5/5 in this department. This shows that the students who did attend the event enjoyed themselves. The students answer the question "about the student's overall experience at ACSTEM?" 81% of the students picked the choice "excellent" which represents 51/63 people which also indicates that students enjoyed themselves at their event

Regarding Improvement for future ACSTEM events, there could possibly be room for improvement regarding sessions about "information applying to Imperial". As 16% of respondents(10 people) pick satisfactory for the session. 16% is still a significant amount of people who feel like they were not overly happy about applying to Imperial through UCAS. Most students would use UCAS to apply to university, and the students that pick satisfactory for this question might not apply to Imperial as maybe they did not receive the information which was needed or relevant information.

In addition, looking at the result from the question "What academic stream did you attend? The two academic streams that had the lowest attendance were science and business. Science stream only had 10% of the respondents attending which is 6 people, business only had 8% of the people attending this stream which was 5 people.

From looking at these results it will probably be best if these two streams are not included in the next conference as it seems the students are not interested in this academic stream. This could allow the opportunity for both of these academic streams to be replaced, or allow more focus on the academic streams that were popular: like Engineering, medicine and computing which all score a total percentage of over 20% of students attending.

To present the long text data I used a world cloud. A world cloud is a great way to represent qualitative data as it displays terms that appear the most within a text

From looking at the world cloud some positive key terms were displayed in the world cloud. Such as the words informative, engaging and exciting. But one of the most shown key terms was "yes". This also represents good feedback from the student's responses toward the event.

Improvement and recommendation

Improvements to the report include the addition of a navigation button such as a button and bookmark. This would allow the user to navigate through the report with a seamless experience and provide a better user experience for the user. Another challenge was presenting the data that had long text data. The only suitable visual was the cloud visual to present the long text data. However, it could be beneficial to see if there are any other visuals which Power BI Premium offers compared to the standard vision I was using.

From looking at the data and results it seems like the event was a success and only slight changes need to be made to further improve the event for next year. Such as changing the Acstem stream structure with different subjects the student would find interesting. For further analysis, it would be really beneficial to compare the result data to future events to see comparison. This could establish if these slight changes would improve the experience for next year for student