Multiple Select Data

Let’s say that you’re analyzing political survey data to understand how to help a particular candidate whose campaign team you are on.

One of the questions is asking the survey participant which political candidates they would vote for. The answers are multi-select, which means that the participants can select as many political candidates as they would like.

What kind of insights could you draw from this dataset?
-------------------------------------------------------------------------------------------------------------------------------------


**clarifying questions:**
over what time period is the survey done? what additional data the survey collects?
where the survey participant is from? what state/region?
do we collect his prev. political prefernces, his age, gender, income other crucial data?



**assessing requirements:**

the insights we derive depend on the features we collect in the survey.


**solution:**

assuming crucial demographics info is gathered on 
each candidate i can build a model that 
predicts (yes/no) or gives score how likely voters
of specific age,gender, location would vote for a candidate or not.

or i can do EDA analysis and see the ground truth - avg age of voters who have specific preferences - left right etc


**validation:**

use std metrics to evaluate the quality of the model
---------------------------------------------------------------------------------------------------------------------------------
Look into which political candidate is popular with which segments of participants eg - age, region, sex etc.

Are participants leaning towards political candidates from maybe Party A or Party B etc.

Which are the user segments that would likely vote for my political candidate? Depending on if a particular user segment doesn’t want to vote for my political candidate, try targeting campaigning to get these voters.

What % of users will vote for my political candidate?

Which candidate seems to be winning according to the survey results?