Analyze Co-Occuring Needs of Inquirers

[Anonymous]

I have been asked to perform analysis on our data & am truly at a loss for how to do it. Background: we operate a helpline where people can ask for various types of community resources. People may ask for one type of need, but many people ask for multiple services in the course of a phone call, chat, etc. We make referrals to organizations based on the person's need & the organizations' capabilities to meet those needs.

I export data in CSV & add it to a folder. When the data refreshes each day, it adds the latest data to the report. The needs data occurs in rows with 1 need/row. The CallReportNum is a unique identifier for the inquirer.

You can see based on the sample that the person asked for help paying an electric bill, their water bill & paying their rent. What I've been asked to do is determine things like: how often do people who ask for help paying rent also ask about help finding employment or help to get food? If someone asks about state benefit programs, what other types of assistance do we find for them?

If the data occurred in columns, I think it would be much easier to determine. Since it exports as rows, however, I am at a loss for what I can do to do the analysis.

Any help along these lines would be greatly appreciated. Since I am not an expert user (and am the most advanced user at my organization), I 

[Anonymous]

@rsbin the dataset usually has 5-10,000 calls/day. The duplicate TN's appear because we may make several referrals to an organization for a need. Each referral has the same TN, but appears in a separate row on the CSV file.

When we count the # of needs, we ask for a distinct count of the CallReportNum column. It is one need with the potential for multiple referrals.

[rsbin]

@Anonymous,

How many unique TaxonomyNames are there?  

As I think about how you can display this, I am picturing an x-y scatter plot   Each Taxonomy Name on both axes.  (a Matrix visual might work as well). Then each data point will represent the number of calls for each combination.  I think this could work if there are a limited number of TN's.  However, this may get ugly if there are too many Names.

Hope you get the idea or at least gets you thinking about other ways to convey the information.

Regards,

[Anonymous]

@rsbin A scatter plot may work.

Thanks for the idea!

[rsbin]

@Anonymous,

I would like to get a better sense of how large your dataset is.

How many unique calls per day?

How may unique TaxonomyNames are there?

And lastly, why do some TN's show duplicates for the same Call Number?