Insights Found By Analyzing My Text Messages

Most of my conversations these days have at least some component that involves text messages. I was curious what I could learn by analyzing my text messages.  How many do I get a day?  From whom?  What are they about?  What time of day?  I put my data skills to work and dug in.  Getting the data from my iPhone and preparing the data was most of the work and a topic for another day.  For now I want to focus on what patterns emerged.

COVID-19 Brought More Than Illness

As COVID-19 came into the US, the volume of text messages I received greatly increased, driven mostly by group text messages from my friends

graph of the number of text messages sent vs. received by month

March saw a 199% month-over-month increase of group text messages compared to February.  This was driven from my friends

Number of Group Text Messages over month

Pandemics Bring Out Potty Mouths

The large increase in group text messages brought a large increase in profanity-laden messages.  I guess my friends have potty mouths especially during a pandemic.

Graph with the Number of Texts with Profanity by Month

11am is When I Sent and Received the Most Texts

Graph showing trend of text messages sent and received by hour

My immediate family was a strong driver of the 11AM texts, especially those I sentDetails of text messages sent and received for the 11AM hour

Thursday and Tuesday Dominate Text Volume, Sunday and Monday Are Quiet

Text Messages Sent and Received by Day of the Week

Under the Same Roof? Texts Still Flow

Being under the same roof doesn't stop my family from communicating with text messages.  In fact, it was the largest group of text messages.  "What's for dinner?  On my way.  Can you get milk? " 

graph showing the number of texts received and sent by contact groups

Does Anybody Really Know What Time it is?

I tapped into the Microsoft Cognitive services to append some additional points to my text data.  Apparently a lot of my messages involve a time component as shown below.  Here are some examples:

Examples - Recognized Date / Time Entities

  • "OK. Well we can fill it out and you can take it in tomorrow if that’s the case."
  • "Okay. I’ll look tonight."
  • "Hey Andy, it's Kevin. 12p is fine."

Examples - Recognized Location Entities

  • "Chris are you still in Dayton?"
  • "Going to Indianapolis for Sarah's band competition"
  • "From a woman living on Bainbridge:"

Table showing the recognized components in my text messages

Ben Folds, Taco Bell and Costco Appear Most

Using Microsoft's text analysis service, I extracted the most common entities.  Ben Folds and Taco Bell?  Sure.  One that is not right is "I can drive". which apparently is a 1995 song.  That is not want i was referring to when I texted that phrase.

> 75% confidence

Stay Tuned For Part 2 - We'll Get Sentimental

In Part 2 of this blog, I explore much more on the content of messages sent including some powerful sentiment analysis powered by Microsoft.  
Graph showing the sentiment analysis of my text messages (positive, neutral, negative and mixed)

Interact With the Dashboard Yourself!

Dying to know the distribution of texts by day from friends Pre and during COVID?   Or what does the trend line look for group messages for extended family?  Get all your answers here! #yourwelcome

Direct Link to Dashboard - Tableau Public 

 

 

2 thoughts on “Insights Found By Analyzing My Text Messages”

Leave a Reply

Your email address will not be published. Required fields are marked *