Enlisting the power of AI to find your 'perfect match' in a coach

Alan Couzens, M.Sc. (Sports Science)

March 27th, 2017

-

In my last post, I talked about the wide range of coaching styles and personalities that you might come across in your search for your perfect ‘coach match’ – from 'the sports scientist', to 'the cheerleader', to 'the boss', to 'the companion'. I also suggested that matching your own personality and needs as an athlete to that of your prospective coach is essential to your mutual happiness and long term success!

I threw out the comment that your best coaching relationships will come when these needs & traits match up ‘ala E-harmony’. And that one little off the cuff remark, got me thinking…

Wouldn’t it be great if we could apply some personality algorithms in the same way that e-harmony does for your romantic relationships to your coach search? Wouldn’t it be great if there was a more general ‘personality matcher’ out there? Well it turns out, silicon valley beat me to the punch again :-)

You might remember the clip from above. Back in 2011, IBM’s Watson AI computing system trounced the reigning Jeopardy champion Ken Jennings. This was a remarkable accomplishment at the time in the field of AI. You might be surprised by that! "Well, of course Watson would win. It just has to “Google” the answer. Right?" Well, first of all, a condition of the challenge was that Watson would ‘stand alone’ & not have access to the internet (in the same way that Jeopardy players aren’t allowed to whip out their smart phone & Google the question :-) but more importantly, even if you were allowed to whip out the smartphone & google the answer, it probably wouldn’t help, right? Because Google would spit out a bunch of keyword search results that need to be waded through in order to come up with a meaningful answer in a timely manner. This ‘meaningful answer in a timely manner’ bit is where Watson shines, in an AI field called NLP – Natural Language Processing, i.e. 'making quick sense of a (huge) bunch of words'.

At the time, Watson was built for the pure sake of pushing AI development, for the ‘challenge’. The next logical step in AI’s progression that followed on from IBM Deep Blue’s super-computer victory over chess grand master, Gary Kasporov. But, as time has gone on & the amount of text based data has grown (twitter, facebook, Wikipedia et al.) & the real world (& commercial) applications of the abilities of something that can do what Watson does, i.e. make sense of HUGE amounts of unstructured text data are becoming obvious.

Sensing this, IBM has given us, the public, access to the extreme power of Watson via the Watson API. Furthermore, they have developed some ‘off the shelf’ applications of Watson’s power and, coincidentally enough, one of them is their ‘Personality Insights’ service.

Here is a description of what the service does from their front page…

If you replace the word “business” with “coach” and the word “client” with “athlete” you start to get a sense of the potential power of this sort of application to our field….

Any coach who's been in the game a while can see the power of the above: A tool that will speed up the process of "getting to know" your athletes - without going through all of the trial and error process of what motivates this athlete - a friendly ear? some tough love? good cop/bad cop? :-) &, just as importantly getting a sense for whether 'this will work' before investments are made and contracts are signed. Fundamentally, both coach and athlete want to find a good match as quickly as possible and Watson can help with this.

So, how do we access the power of Watson? All it takes is a few lines of code…

A quick rundown on the code…

First we import the required modules for the analysis from Watson & Twitter, along with a ‘pretty print’ library that will print out the mass of json data that Watson returns in a nicer, more decipherable format. We then give the script the twitter handles that we want to analyze (here it’s blank, but you’d replace in your script with the twitter handles of the people you wanted to analyze. e.g. handles = [‘alan_couzens’,’bob_smith’]. Then we provide your developer keys from Twitter and Watson. Sign up for those here & here. We get a ‘text’ readout of each person’s twitter history for analysis. Then we run personality_insights.profile on the text – telling Watson to analyze this mass of twitter text for key personality insights of the coach and athlete. This will spit out a huge amount of data, so we drill down deeper into one of the measures that Watson provides – the ‘big 5’ personality inventory. Then we compare the athlete’s and coach’s responses along each of the 5 measures – openness, conscientiousness, extraversion, agreeableness & neuroticism to see how they match up. Eazy peazy.

By running the above on yourself and a prospective athlete (or yourself and a prospective coach!), Watson will analyze your respective social media feeds, it will pull key personality words out & plug them into its validated machine learning algorithms to spit back a whole bunch of comparative personality data, which you can then compare against each other to see if you're a 'match'…

For example, the below data compares myself (blue bars ) with an athlete (red bars) that I've worked with for a long time, had a good deal of success with &, really enjoyed every step along the way. Watson assesses each of us for the 'Big 5' personality traits - openness, conscientiousness, extraversion, agreeableness, and neuroticism and returns a value of where we rank across the sample mean, i.e. a score of 80% means that we rank higher on this trait than 80% of people sampled. You can read more about the ins and outs of each of the personality scales here. But below is a quick synopsis of each...

Openness, or Open to Experience, is the extent to which a person is open to experiencing a variety of activities. It includes things like adventurous, artistic, creative, curious, authority-challenging

Conscientiousness is a person's tendency to act in an organized or thoughtful way. It includes things like achievement-driven, dutiful, organized, committed, persistent

Extraversion is a person's tendency to seek stimulation in the company of others. It includes things like - energy, assertiveness, cheerfulness, friendliness

Agreeableness is a person's tendency to be compassionate and cooperative toward others. It includes traits like altruism, compassion, sympathy, modesty, trust.

Neuroticism also called Emotional Range or Natural Reactions, is the extent to which a person's emotions are sensitive to the environment around them. It includes tendencies toward anger, anxiety, depression & self-consciousness

After pulling these values for coach and athlete, the script compares them against each other as shown below...

Overall Coach Athlete Match:

Openness:

Conscientiousness:

Extraversion

Agreeableness

Neuroticism

The above match can be classified as a very productive one. So, given this insight provided by our good friend, Watson, when I come across future prospective athletes with these traits, I will be apt to take note of that (and, yes, if you're a prospective athlete, you can rest assured that I'll be running these 'background checks' before working with you :-)

On the flipside, I have also run the same analysis, retrospectively, on previous relationships that weren’t as compatible. You can click 'bad match' above (or below :-) to see the difference....

:-) Sure enough, the percentage match is considerably lower - typically less than 50%. You can see the key differences. My ideal athlete is conscientious, sufficiently extraverted to communicate when required, not all that open to new things (new programs, new coaches etc. :-) & committed to stay on the path for as long as it takes, agreeable enough to do the work but disagreeable enough to be a challenge & finally, just a bit neurotic - dissatisfied enough with where they currently are to be motivated to do what it takes to get 'there'. Or, in the words of Ray Lewis - "pissed off for greatness".

The above analysis only scratches the surface of the data that Watson provides. Much of the deeper benefit extends beyond a pure 'match'. Watson also provides a ‘needs analysis’ that goes beyond basic personality traits to look at what elements of the relationship are likely to be most important to the respective parties – things like ‘closeness’, ‘stability’, ‘structure’ - again things that are very applicable to a coach-athlete relationship! And things that tie in very closely with some of the skill-sets of different personality types that I outlined in last weeks post. I’ll talk some more about how these additional key insights can help a coach in tailoring their communication to the specific needs of the athlete & and how they can help the athlete in communicating to ensure they get their own needs met in a future post…

Until then….

Train smart,

AC

  Tweet