Confusion Matrix with Johari Window

I was discussing Confusion Matrix with my friend and then my wife who is an Industrial & Organizational Psychologist asked me about it. I tried explaining her in terms of data science however I could realize that she was not able to understand it so I went ahead explaining it in her known terms.

There is a concept of Johari Window which is mostly used by HR people and Psychologists.

The Johari window concept talks about the things which we know / don’t know about ourselves and similarly others know/don’t know about ourselves.

So, I drew parallel between the two, Confusion Matrix and Johari Window to make it easier to understand.

I asked my wife to assume that you are a model. There was an instant response from her ‘What do you mean by assume? I am a model’ (I was happy with this confusion ?? )

So there goes the below conversation with her i.e., with my model.

Me: I will share 2 things with you about myself and you have to tell me whether you knew it or not

Statement1: I am good at data analysis.

Model: Yes, I know

Statement2: I am doing a unique project in NLP.

Model: I didn’t know this

Model: But I know you can do far better than what you are doing currently

Me: Thanks, I didn’t know this

Me: Will I be leading an AI company in future?

Model: I don’t know this currently

……

And here I stopped this conversation with my ‘model’.

Its time to get into the explanation of these 2 concepts as a single concept.

So, I draw below 4 quadrants

I then asked my model to put the above 4 statements in the related quadrant and model did

Seeing this, my wife said ooh It’s a Johari Window. I said Yes, its Confusion Matrix too.

So, assume there are these 2 classes ‘Known’ and ‘Unknown’

If the ML model predicts the class ‘Known’ as ‘Known’ then it’s True Known or True Positive

If the ML model predicts the class ‘Unknown’ as ‘Unknown’ then it’s True Unknown or True Negative

If the ML model predicts the class ‘Known’ as ‘Unknown’ then it’s False Unknown or False Negative

If the ML model predicts the class ‘Unknown’ as ‘Known’ then it’s False Known or False Positive

Now let’s put these 4 results in the same Johari Window.

So here is my combined explanation to her.

And as expected being a psychologist she asked me what do you do with this confusion matrix? I knew from where she is coming. In Johari window there is an actionable for each quadrant. Its good to take feedback for FP cases so this FP becomes TP later. Similarly, in ML algorithm we do give feedback to our model which is called as ‘Hyperparameter tuning’. This is to ensure that the model has better predictions of classes.

Now she is clear on Confusion Matrix and I hope you too………??