Topic 8 Catch up Day

Learning Goals

• Give and receive constructive feedback on writing about data and machine learning concepts
• Synthesize and apply concepts covered so far on new project data

Slides from today are available here.

Content Conversation 1 Prompts

In preparation for your content conversations next week, please go to the Google Doc here for all of the details and instructions.

Here is a brief summary.

• You’ll meet with me in your final project groups
• You’ll need to schedule a meeting with me by choosing a time slot on my calendar (link in the Google Doc)
• You can meet with your group ahead of time to practice/prep for this conversation
• Prompts for the conversation are available in the Google Doc.

Peer Review

In groups of 2, you’ll do the following:

• Send the link to your portfolio to your groupmate
• Choose 2 concepts/topics that you’d like feedback on
• Think about what aspects of the writing you’d like feedback on (e.g. conceptual, clarity, creativity)
• Tell your groupmate the concepts/topics and aspects you’d like feedback on.

Spend 10 minutes reading your groupmate’s writing.

• Make comments in the Google Doc (focusing on the aspects that your groupmate wanted you to look at)
  • Comment on sentences that you really like and why.
  • Comment on sentences that are not clear.
  • Comment on sentences that suggest a misunderstanding.
• If your groupmate is open to you providing wording suggestions, change from “editing” to “suggesting” and offer an alternative way to write something.

Spend 10 minutes discussing the comments for one of your writing and then switch.

Project Work

By the end of today, your group should

• finalize the data set you’d like to work with
• make sure that you can read it into R
• made some initial progress on visualizing and cleaning the data
  • choose on quantitative outcome variable for regression
  • focus on a subset of about 10-20 predictor variables for now
    • for some of you, that may mean that you need to create new variables out of existing ones (like time/date)
    • for some of you, that may mean choose some from over 100 variables
  • focus on a subset of rows that you can easily load into R
    • if the data won’t load in quickly, focus on a subset of the data (by year or another characteristic) and then create a test set and a training set (save them into csv separately using write_csv)
  • look for missing data & outliers
  • understand what the values of the variables mean
  • get a sense for possible relationships that may exist in the data