Okay, it's recording.

Today's lecture is about correlation and regression.

The agenda is pretty simple. First we start off of correlation and then we start

talking about regression as suggested in a rule title of the lecture.

First simple linear regression and then we will talk a little bit about multiple

regression. The purpose of this lecture is to give you an intuition about the

concepts behind correlation and regression and also point you to some of the

methods and the concepts used in calculation and correlations and

progressions. So let's start with correlation. I am pretty sure that all of you

have seen a visualized correlation since one way or another. There are two

types of information that correlations convey to you. One is the direction of

a relationship as you can see with the upper three plots with the green dots.

It can be a positive correlation, it can be a negative correlation and it can

be a null correlation that is no correlation. In addition to the direction of

the relationship, you get information about the strength of the correlation and

in visualized means the strength is communicated by the spread of the data

points. So each data point gives an information about two variables that are

measured in relation to each other. In the first of the lower plots you see a

strong correlation because they are the spread of the data points. It's

relatively low. Then in the middle you see a weak correlation with a larger

spread and then the last plot is a new strength like again a null correlation.

Excuse me, I'm sorry to interrupt but we can't see any slides on the team.

You can't see the slides. Thank you for notifying me. You can't see any slides at all or do you see?

No, we can't. We just see you. That's it. The webcam. Okay.

Do you see the slides now? Yes, let me see it now. Thank you. Good.

Do you see them also in the right presentation mode? Here you see the plots to

what I just told you. I'm pretty sure that you've seen similar plots.

This is just to give us a start and introduction into the concept of correlation.

Importantly, correlation while describing a relationship between variables,

correlations do not need to be causal. A famous example to make this point that

a correlation does not have to be causal is a strong correlation that was found between

the number of breeding stalks and the number of newborn babies in Eastern Germany

between the years of 1966 and the 1980s. I don't know if you're familiar with this legend,

the story that people used to or still tell children when they asked where babies come from,

they tell them that in Western countries they tell them that stalks bring babies

and there is in fact, or there was in fact, quite a strong correlation between the number

of stalks and babies, which you can argue, yes, well, that is evidence for this legend

that is told that stalks in fact bring babies because when they're less stalks, then obviously

we will get less babies. But this is a fallacy. It is a fallacy because there is no causal

relationship as I'm sure that all of you know babies do not are not brought by stalks.

And there are other variables that can affect both the population of stalks and the birth

rate of babies. In this case, one explanation that is given for this relationship is that

this was a time the 1960s to the 1980s in Germany of industrialization, which essentially

meant that living in rural areas became dirtier. There were some more emissions and these emissions

affected the survival of stalks and at the same time and independently from other industrialization

also affected the birth rate of babies because for example, more women went into the work

for more families decided to have less children. But you have to be aware that you can or when

you bring variables in a relationship in the graph, then you are transporting the message

that there may be a causal relationship. So you have to be very mindful of inducing what

correlations you are actually presenting and then also how to present them. For example,