A2 - Group Report [CLO1, CLO2, CLO3, CLO4]
Week 7 (Sunday 27th October) 11.55 PM / Weighting: 30% / Length: Max 4,000 Words (+/- 10%)
Description
This assessment provides you with the opportunity to apply the concepts learned in class to develop and communicate an appropriate image advertising strategy.
Your team is working for an advertising agency company. One client just launched its business in the X industry. The client requires information on what effective visual image advertising content strategies in the respective industry look like.
Details
You will be collaborating with your peers in a group of 4 to 5 students to complete this task. Follow the below steps to complete your task:
1) Data: You will receive data files (in your Assignment 2 folder) which will contain meta. data, images and selected googlevision files from scraping and analysing Instragram posts for Lorna Jane (https://www.lornajane.com.au). Lorna Jane describes itself as an active living brand which seeks to empower women and transform. the world.
All images in the image file are photos.
General: For Part A, you will need to run Open CV on the images, sentiment on descriptions contained in the metadata and merge the relevant data files to produce a suitable regression file. This regression file must include control variables for time.
For Part B, you will need to analyse the googlevision data files provided and merge the created data files with other relevant files to produce a suitable regression file. This regression file must include control variables for time.
For the data analyses you conduct be mindful of the potential need to select sub-sets of the data depending on trends over time and/or outliers, both of which may impact on results. For regressions, ensure your final results are not unduly impacted by multicollinearity.
PART A:
For this part, use the OpenCV, PB and Text sentiment (description) data files and/or merged data files (including control variables for time)
2) Hypotheses: Advertising professionals often share effective visual content from their industry experience, although those are not formally tested yet.
Search for recent articles such as academic journal, newspapers, blogs, and industry reports about effective content strategy on social media. From your search, formulate some preliminary hypotheses about main effects (use only OpenCV features and/or description text features) and possible moderator effects associated with these main effects.
Discuss the reasons for your hypotheses by citing articles which you found. The main effects hypotheses don’t need to be related to each other but the moderation effects hypotheses should be connected to one of the main effects hypotheses.
From your research provide possible candidate hypotheses (< 6) about main effects of either visual or text features of images (OpenCV features, text features) and about moderation (< 4) of the chosen main effects. You may not find specific research related to lifestyle. brands so justify why your research/hypotheses may be applicable.
3) Data-Analytics-Evidence:
For this task, you will use descriptive evidence such as summary statistics and plots to provide evidence (or otherwise) for your preliminary hypotheses or to discover other relevant hypotheses which appear evident from the analyses.
For example, look at summary statistics and/or make an XY plot for each of the preliminary hypotheses about the main effect (X = a variable corresponding to your hypotheses, Y = the log (like count + 1)). For moderators, make XY plots conditioned on ranges (eg. High, Medium, Low) of values of the main variable.
The above analyses and visual patterns may help you choose your final hypotheses among all the preliminary candidates. Specifically, discuss why you kept, dropped, or added hypotheses to your final set compared to your preliminary research.
4) Final Hypotheses: From the above final set, you should now choose three Hypotheses about main effects and two hypotheses about moderators (must be connected to the main effects hypotheses).
Explain your choice of final Hypotheses logically (eg., why does your X increase Y?) by citing the above analyses, related theories and any other supporting evidence.
5) Testing your Hypotheses:
(a) State your regression model with the definition of each variable and coefficient. X variables should include the variables connected to your hypotheses and suitable control variables including posting time dummies: Year, Month-of-Year (January, …, December), Day-of-Week (Monday, ..., Sunday), Time-of-day (Morning, Afternoon, Evening, Night).
(b) Run the relevant regression considering potential multicollinearity issues. Report the regression results in a suitable Table (make a summary Table that includes only the relevant X variables and their coefficients from the above regression results). Put the full regression results in the appendix. Add colour to significant coefficients to highlight them. Interpret the regression results.
(c) Conclusions: Provide conclusions for your hypothesis tests and make recommendations for Lorna Jane marketing managers based on your insights and conclusions.
PART B:
For this part, use the OpenCV, PB and googlevision files and/or merged data files (including control variables for time)
6) Quick Hypotheses: Consulting companies often need to tell their Quick Hypotheses to clients. Find and report all the top 20 posts vs the worst 20 posts regarding “like count” for all the relevant posts. (Filter out the high outliers)
Compare the top and worst 20 posts using the result of Google Vision (multi-objects, labels, logos, OCR) using its Demo or API. Comment on whether the Google Vision evidence suggests any potential hypotheses that could be considered.
For this section only consider main effects hypotheses. For example, you may discover the top 20 posts are predominantly of persons rather than objects or they contain the brand logo compared to the worst posts. This might suggest the hypothesis that images with persons are more likely to engage the viewers (likes) than images without persons.
7) Using the relevant regression data which incorporates OpenCv, metdata, googlevision and time variables test which key objects/labels (persons, product etc) are best (if at all) at increasing viewer engagement (likes). Provide relevant hypotheses and test these hypotheses. Provide conclusions and recommendations for Lorna Jane Instagram posts.
8) Lorna Jane sponsors the Queensland Firebirds netball team. It often includes the logo of the Firebirds in its Instgram photos and posts. It also includes its own Logo in selected images within its Instagram posts. Provide and test relevant hypotheses (separately) that the Firebirds and Lorna Jane logos impact on viewer engagement.
9) Robustness test: Repeat the above analyses [(7) and (8)] with Y = the log (comment count + 1). Interpret results and discuss whether the results in [(7) and (8)] are robust.
10) Conclusion:
(a) Deliver your findings to Lorna Jane by recommending effective visual image advertising content strategies with supporting evidence.
(b) Also, as one way of communicating your findings, generate TWO new advertising prototypes based on your recommendations (for either Part A, Part B or both) to increase viewer engagement. Advertising agencies often give two versions of ad prototypes to their clients. It does not have to be a fancy ad. For example, if green turns out to be the best colour, you can use the green colour in your prototype.
There are many online resources that you can use including CANVA which has many templates.
● https://www.canva.com/templates/search/instagram-posts/
● https://www.canva.com/create/instagram-stories/
Report Details:
In completing the tasks, apply appropriate data analytics and consider the concepts introduced in class. Your report should not exceed the word limit, excluding the title page, relevant images, tables or charts.
Title page (1 page) includes (1) The title of your report, (2) The word count, (3) An executive summary (One paragraph) of your report, (4) the course name, tutorial session and group, tutor’s name, (5) Each team member’s first and last name & zID
References: Cite academic papers, newspaper articles, blogs, or industry reports using Endnote. Use APA (American Psychological Association) style. in-text citations and a reference list at the end. https://student.unsw.edu.au/apa
Appendix: For the top 20 posts vs the worst 20 posts (screenshots of both image and text description part) regarding “like count”, report at most 4 posts (the most relevant) for each.
Format: Use word file (.doc), 12pt, 1.5 lines spacing, at least 2.5cm margins on all sides.
Submission instructions
A. Submit your report (only once per group) to Moodle Report submission box
File name: Tutorial_Group_A2.doc” (e.g., W12_1_A2.doc)
Ensure you provide the names of the group members on the cover sheet/front cover.
B. Submit other supporting files (data, image, paper and code) to the Moodle Other Files - submission box. You can, in addition, put all your supporting files in a link to a G-drive folder which
1) .xlsx file contains the datasets on which you run the regression.
2) .ipynb contains all relevant code to get the results in your report. Make a zip file by combining all colab files.
3) .xlsx also contains all the cited paper lists with a brief note about why you cited them.
4) You can also provide a document with a link to a shared G-drive folder containing all your group’s images and relevant files (ensure the link works)
● For each missing file among the above (1) to (4), -1 mark
Marking Criteria
Your assignment will be marked based on the following marking criteria:
1. Analysis: Quality of advertising image data analytics
2. Hypotheses: Quality of Hypothesis development
3. Written Presentation: Quality of written report
For further information, see the below marking rubric.
Marking Rubric for Assessment 2: Industry Benchmarking – Group Report
|
Criteria
|
%
|
Fail
|
Pass
|
Credit
|
Distinction
|
High Distinction
|
|
Analysis
Quality of advertising image data analytics
|
30%
|
Analysis does not meet the required standard.
|
Sufficient analysis of the advertising data, which identifies and measures Instagram data. Some attempts to do regression analysis and interpretation of results with some accuracy.
|
Proper analysis of the advertising data, which mostly identities and accurately measures posted Instagram data and presents findings. Attempts regression analysis and includes some X variables, addressing multicollinearity issues. Interprets results to some extent.
|
Effective and proper analysis of the advertising data, which accurately identifies, and measures posted Instagram data, and presents findings clearly in an appropriate format. Does regression analysis properly by including necessary X variables, addressing multicollinearity issues, and interpreting results to an appropriate standard.
|
Highly effective and proper analysis of the advertising data, which accurately identifies, measures, and compares posted Instagram data; clearly and accurately presents findings in an appropriate format.
Does regression analysis properly by including necessary X variables, addressing multicollinearity issues, and interpreting the results properly.
|
|
Hypothesis
Quality of hypothesis development
|
50%
|
Hypotheses development does not meet the required standard.
|
Sufficient development of hypotheses by applying most of the required steps and addressing most of the relevant features; attempt to make final Hypothesis with some discussion provided.
At least one Hypothesis needs to be significant. Findings are most appropriately communicated. Attempt to provide recommendations.
|
Good development of hypotheses by applying required steps and addressing relevant visual features; the final Hypothesis is appropriate; discussion is supported by considering previous data and results.
At least one Hypothesis needs to be significant. Findings are appropriately communicated, and some recommendations are provided.
|
Effective development of hypotheses by accurately applying required and relevant steps and addressing relevant visual features; the final Hypothesis is appropriate and evidence-based; discussion is supported by considering previous data and results and concepts.
At least two hypotheses need to be significant. Findings are clearly communicated by recommending image advertising content strategies.
|
Highly effective development of hypotheses by accurately applying the required and relevant steps and addressing relevant visual features; final Hypothesis is highly appropriate, meaningful and evidence-based; discussion is supported by considering previous data and results, concepts, and scholarly articles.
At least two hypotheses including one main effect need to be significant. New insight is made. Findings are effectively communicated by recommending effective image advertising content strategies with supporting evidence.
|
|
Written Presentation
Quality of written report
|
20%
|
Report lacks clear structure. Written English is below the required standard.
|
Report provides a mostly appropriate structure with distinguishable paragraphs. Written English is appropriate to the task but has spelling, referencing and/or grammatical errors.
|
Report is clearly structured with good transitions and paragraphs. Good use of written English, which is appropriate to the task and has few spelling, referencing and/or grammatical errors. Report mostly adhered to the prescribed word count and conventions.
|
Report is clearly structured with excellent transitions. Above standard use of written English language, which is professional and appropriate to the task with minimal spelling, referencing and/or grammatical errors. Report adheres to the prescribed word count and conventions.
|
Report is clearly and logically structured with excellent transitions and paragraphs. Excellent and above standard use of written English language, which is professional and appropriate to the task and has no spelling, referencing and/or grammatical errors. Report adheres to the prescribed word count and conventions.
|