Update 🏠Home.py

🏠Home.py

@@ -33,6 +33,8 @@ st.markdown('''As machine learning techniques advance, algorithmic systems play
 
 st.markdown('''Defining the target variable is difficult. Imagine that you are hiring a salesperson. What makes for a good salesperson? Simply someone who makes the most profitable sales? Or is a good salesperson also a good leader? Does a good salesperson come up with new ideas that can improve how the sales team operates as a whole, and not just their individual sales? (The list could go on.) Perhaps the answer is: some of everything.''')
 
+st.video("https://www.youtube.com/watch?v=ck1hRTkUtsM")
+
 st.markdown('''But then we ask: how much of each thing? How much more important are individual sales than leadership, for example? Put another way: there may be different ways of understanding which qualities matter for being a good salesperson, and to what degree; reasonable minds may disagree on these issues (as anyone who’s been on a hiring committee has experienced). Even once it’s decided what makes for a good salesperson, there is a further question of how to make the notion precise in algorithmic terms: how do we identify job applicants with sales ability, leadership qualities, or innovative thinking? In order for the algorithm to be able to positively select those applicants, those qualities have to somehow be encoded numerically.''')
 
 st.markdown('''Defining the target variable is not only difficult; it can also have profound effects on fairness—by resulting in hiring disparities for protected groups [(Passi & Barocas 2019)](https://dl.acm.org/doi/10.1145/3287560.3287567). For example, if you define the notion of a “good” employee in one way you might end up hiring more women than if you were to define “good” in another way. Relatedly, machine learning models might behave differently depending on how “good” employee is defined. Defining the notion in one way might lead to your model being less accurate for older applicants than for younger applicants.''')