Memo 1 Writing Research Proposal Memo Overview In this assignment you are going to write a 2 page memo by filling out the various pieces. This memo format is based on a common format in industry. The folder with all Memo 1 materials including the assignment, rubric, and examples is located hereLinks to an external site. The template for Memo 1: see attached doc The rubric for Memo 1 can be found: see attached doc Examples can be found: Memo 1 Example 1Links to an external site. see attached doc You will be writing a 2-page Memo. If you go over 2-pages it needs to be in the appendix and referenced in the other text. This should never go over 5 pages! The Scenario: You are a Data Scientist working for a Tech Company. You have been asked to understand the effect of COVID-19 on retention. You need to write up a Business/Research plan for explaining the growth or loss of retention on the platform (i.e., THE APP). Key things to consider: How do we define retention? First we need to define active users. We will use 30 day active (e.g., the account has been used in the last month). If the account was used in the last 30 days then the user is considered active. Growth would be more users active over the last 30 days from day 1 to day 2. Loss would be fewer users active on day 1 to day 2. How do we model (perform research) on retention? Here we will use either a linear or non-linear growth model. The key question for you is what variables you would use to model the growth (positive or negative): E.g., gender This is limited to what is readily accessible to the COMPANY. Other considerations: Do you need measures on the users directly beyond TRACE DATA. E.g., Would you want a survey as part of your research plan, if so why? What would the survey look like? What questions would you ask? What you need to do: Start by selecting a Tech company that you are writing this for: Facebook, Inc Facebook WhatsApp Instagram Messenger Amazon Apple Netflix Google Microsoft Other Tech Companies not on this list: You can choose another company, but you need to write a paragraph about the company. Provide basic information such as size of the company, major products, what their market is etc. If you choose a company not on the list, you cannot assume I or the TAs know anything about the company so you need to be clear that all necessary information needed to evaluate your report is written up. Memo components: Come up with a TITLE. TL;DR Too Long Didnt Read is 3 one line sentences (no returns) that sum up the document for a Director or VP who has maybe 30 seconds for your memo. Write this last. Business Case This paragraph answers the question of why we are doing this analysis/engineering/research/etc. I.e., What is the motivation for this proposal that is easy, simple and motivating. E.g., Why does the business care about this project. Problem/Research question what is the problem or research you are conducting and basics on why it is interesting. This should be a question. Hypothesis(es) What you think the possible outcomes will be. Usually written up with Null hypothesis and one or two alternative hypotheses. Analytic Plan This is what steps you are going to take to solve this problem. In this case you are going to use a growth model (you need to explain why you would use a linear or non-linear growthLinks to an external site. model). The big exercise here is to explain what variables your model would take (e.g. gender, age, etc). You need to mention whether the variable is available (e.g., the company has a log of the variable created by user activity) or whether it needs to be collected (e.g., survey) or purchased (e.g., third party vendor). If it needs a survey or needs to be purchase a more detailed explanation is required. Linear Model Yt=b0+b1X1+ +bkXk where Yt is outcome variable of interest and Xs are the covariates you choose. Exponential Growth model can be linearized and written as follows: g(Yt) = b0+b1X1+ +bkXk Same logic the key variable needs to be selected and why. If they are not available in the log data then collection needs to be justified. A discussion on how to think about modelsLinks to an external site. is written up See these links from Khan Academy Linear vs. exponential growth: from data (example 1)Links to an external site. Linear vs. exponential growth: from data (example 2)Links to an external site. You do NOT need any mathematical equations in your write up, but you should be able to articulate how you think the relationship between the outcome (e.g. retention) and your metric for COVID-19 (e.g., number of cases) relates (e.g., linear). Timeline A simple spreadsheet of all items that need to be accomplished and how long you expect to take. For this scenario you have 2 weeks to accomplish the analysis task. Make a timeline that gets everything in two weeks. Be creative. https://docs.google.com/spreadsheets/d/1EfHcq7mEopjpQ5yI4EtiL0bSRZoAydxm/edit#gid=353239173 Example Timeline can be found: see attachments You should use real dates (e.g., 06/10/2020). Using a Gantt chartLinks to an external site. can be visually helpful for understanding how longs take and where actions overlap. Appendix Any extra details or pre-analysis goes here. This should occur on page 3 and should only be used if needed.
