代写Brand Analysis、Moodle Site代做、代写Python、Python程序调试

日期：2019-05-12 09:44

ANALYTICS SPECIALIZATIONS & APPLICATIONS, 2018/2019

COURSEWORK 2 - Brand Analysis using Twitter

Final Report Deadline: Wednesday 15th May 2019, 3.00pm

Submission: Electronic submission via the module’s Moodle Site.

1. The Problem Definition:

In this coursework, you have been asked by a company to do an exploratory “snapshot” analysis

of the public’s view of a specific brand/product - as seen through the lens of social media (twitter).

This is because the company itself are going to being producing related/complementary products

themselves, and so wish to undertake this first exploratory analysis. You are tasked with:

1. Extracting tweets referring to that brand/product over a maximum of a one week period.

2. Unpacking those tweets’ contents, and from these providing a breakdown analysis of the

brand’s relationship with people on social media (n.b. this might include an exploration of

items drawn from: prevalence of mentions; engagement; key descriptive words occurring in

relation to the brand; topics arising; attitudes or types of users mentioning the brand;

analysis of sentiment surrounding the tweets; geospatial location of mentions; companies

promoting the brand; temporal nature of mentions; etc. - the choice is open and yours, and

you are not expected to cover all of these in a pilot, exploratory analysis).

3. Identifying a candidate micro-influencer in Twitter who you believe may be good for the

company to engage with in the next stage of the pilot..

4. Completing a report (maximum 8 pages, or 3000 words, whichever limit you reach first,

including all visualizations you select - no appendices will be accepted) summarizing these

analyses.

N.b. Importantly you are free to choose the Brand/Product that you would like to examine.

Nonetheless, you absolutely must select a well known brand or products (this might be, but

certainly not constrained to, a leading makeup brand or product, a car brand, a sporting team

brands, etc. - please select carefully, as it must be a brand/product that generates tweets) It would

also be preferable if this brand has a competitor that you may wish also to investigate/contrast

with in parallel (this is not mandatory but may well contextualise your results).

Note also this is a ‘snapshot’ pilot analysis - not a full in depth study!

Inevitably some of your analysis will be more qualitative than if you had tens of thousands of tweets

to analyse (you are unlikely to be able to obtain data anywhere near that amount, nor are expected

to). Similarly you are not expected to find a perfect-micro-influencer - merely someone who is a

justifiable starting point based on metrics that you specify.

Picking a brand/product that will provide you with enough tweets to run analytics techniques on is

crucial - quantitative analysis is mandatory to undertake in some form. Those techniques can

involve any analytics approaches you have learnt and would like to apply (whether it be geospatial,

text analysis, sentiment analysis, time series analysis, etc. - or any combination).

2. Expected Approach:

● It is expected that your data collection approach will use a python script (much in the same

way as the practical from “ASA Week 8 - case study exercises”). This is a key part of the

assessment. If you want to source additional tweets outside of those collected directly by

your python scripts, you must get clearance to do this from myself (and under absolutely no

circumstance may these be purchased).

● Tweets, once obtained by your scripts, must be saved in an appropriate form (you will be

required to submit your data as part of your submission).

● Once generated, that data may then be analysed in any form you see fit - you may analyse

it to evidence reach, mentions, attitudes to the product, sentiment associated with product,

geographical patterns, indeed anything you think may be relevant to understanding the

brand. However, as detailed above, you must apply some form of analytics technique

during this process, rather than taking a purely qualitative approach.

● You may also augment your analysis scripts, and presentation, using information from 3rd

party online tools for your chosen keywords if you so desire. I don’t recommend this, or

spending too much time here, as 3rd party tool use will not be credited in your functionality

marks, and will only contribute to your presentation and professionalism marks. Any 3rd

party tools used in this fashion must also be free and open to use.

● You are expected to find an appropriate micro-influencer by selecting some metrics that

support your choice, and also obtaining these via the twitter api.

3. Report Structure

To present your exploration, you will provide a report of this snapshot pilot study that clearly

describes your target brand product, any competitors chosen for comparison, the approach and

techniques you have taken to underpin your work, and the results of that analysis (including

visualizations where appropriate). You must also identify at least one potential candidate

micro-influencer for the company to engage with (along with some form of justification), before

wrapping up your insights in a final conclusion section. Expected sections are as follows:

1. Executive Summary: including a description of the task, a summary of your technical

approach, a summary of the data that underpins it, a summary of the results, and a

summary of the insights you have arrived at.

2. Approach breakdown: a summary of the process that you have undertaken to obtain your

social media data, to analyse it, to summarize results, and to draw conclusions.

3. Data Collection section: A summary section quickly detailing the data you have obtained,

and upon which your analysis is based. This should include at the absolute minimum

including information the number of tweets obtained, your search terms, the number of

unique users analysed, the date range and the geographical area focussed on (this may be

global, but please do specify).

4. Analysis section: In this section, which will comprise the bulk of your report, you must

summarize your investigations. Please feel free to split this into different subsections based

on the techniques you examined, or angles you took to considering the text surrounding

tweets, the locations they were produced from, and the people/companies that were

mentioning them.

5. Micro-influencer Recommendation: Here you will provide a summary of the twitter user you

recommend to the company to engage with concerning the product.

6. Conclusion: A brief conclusion summarising the key parts of your analysis, and any

recommendations you have for the business if they were to extend this pilot study into a full

analysis.

4. Marking Criteria

Your submission will be assessed based on the following mark scheme:

● Executive Summary (5 marks)

● Methodology Section (5 marks)

● Data Collection Section (10 marks)

● Analysis and Description of Results (50 marks)

○ Brand Exploration (35 marks)

○ Micro-influencer Recommendation (15 marks)

● Conclusion, Insights and Recommendations (5 marks)

● Technical Implementation (20 marks)

- Functionality (15 marks)

- Clarity and Commenting (5 marks)

● Overall presentation and professionalism of Report (5 marks)

5. What you need to submit

● A report on your exploration - 8 pages maximum and 3000 words maximum - whichever

you reach first. See the report section for the structure and to understand what is expected.

● All code / notebooks / tableau files. I expect:

○ A commented Jupyter notebook containing your data collection component.

○ A different commented Jupyter notebook containing the code that reloads in these

tweets, and performs analysis on them that you then describe in your report.

● A supporting data file should be included containing the tweets that you harvested, that

can be run through your analysis script.

6. How to submit

● All files must be zipped into one file. Submission is electronically via Moodle. The submission

link will be made available shortly. DUE DATE: Wednesday 15th May 2019, 3.00pm UK time.

7. Tips

● Pick a brand/product that you find interesting, or might be relevant to your future goals.

● Pick a brand/product that invokes discussion - you are going to have an easier life if your

selection is one that generates lots of tweets (in fact this is crucial)!

● To obtain tweets you can reference the exercises in our social media session - “ASA Week 8

- case study exercises”, or use examples that will be added to moodle using an alternative

library such as tweepy. You don’t have to use this approach or library but it is a good option.

● To analyse tweets you might want to do some simple counts, word counts, etc. You may

also find some inspiration from the example case studies exercises in - “ASA Week 04 -

Text analytics” and “ASA Week 05 - Topic Modelling”.

● Store your tweets into a data structure that is easy to save to file. You might use a pandas

dataframe, the raw dicts that the python twitter library we’ve used in the past returns, or

your own structure saved as a csv. Whatever you find easiest to work with, and load into.

● You may download tweets gradually and piece them together into a single file. This is

perfectly fine - scripts often timeout, and you will gradually expand your input information.

● Start downloading tweets as soon as possible. This will be a bottleneck otherwise.

● Start your analysis approach on a small amount of tweets, and just get something working -

you can always rerun once all your tweets have been collected.

● Don’t try and do to much - think what would be a neat, contained and informative pilot

summary. This is after all, only a ‘snapshot’ (and equally your micro-influencer chosen just

needs to be justified and reasonable)