0
$\begingroup$

Sorry, maybe this is a bit of a rookie question, but I would like to find out the probability of A(tweet sentiment being negative or positive) based B (the length of the tweet).

This to me sounds like a conditional probability problem and know that the formula translates to P(A|B) = P(A and B)/P(B), but I have no idea how to implement in on my R dataframe. Just to be clear I am trying to calculate the probability of a sentiment being positive or negative if it contains over 35 words

This is a sample of my dataframe

Data

Sorry for not producing an example of what I have done, kind of stumped on what to do

$\endgroup$
0
$\begingroup$

In order to calculate $P(A|B)$ using a dataframe in R, you'll want to find the proportion of cases where $A$ is true and $B$ is also true and then divide by the number of observations where $B$ is true. So for example, let's say we have this dataset:

> ##Generate some sample data
> set.seed(3434)
> n<-100
> A<-rbinom(n, 1, .25)
> B<-rbinom(n, 1, 0.6)
> mydata<-data.frame(A=A, B=B)
> rm("A", "B")
> head(mydata)
  A B
1 0 1
2 1 0
3 0 0
4 0 1
5 1 0
6 0 0

Then you can calculate $P(A|B)$ in one of two ways:

Method 1

You could use the formula you mentioned in your post and obtain ${P(AB)\over{P(B)}}$ by summing all the observations where both $A$ and $B$ are true (or equal 1 as I'm assuming 1=TRUE here) and dividing by the total number of observations in the dataframe ($n=100$) to obtain the proportion or probability $P(AB)$, and then dividing by the proportion of those cases where $B$ is true to obtain $P(B)$.

> ##Method 1
> A_true_count <- sum(mydata$A)
> B_true_count <- sum(mydata$B)
> A_and_B_true_count <- sum(mydata$A + mydata$B>1)
> P_of_A <- A_true_count/n
> P_of_B <- B_true_count/n
> P_of_A_and_B <- A_and_B_true_count/n
> P_of_A_given_B_m1 <- P_of_A_and_B/P_of_B
> P_of_A_given_B_m1
[1] 0.2807018

Method 2

You could obtain $P(A|B)$ directly by considering $P(A|B)$ as a reduced sample space. So you simply determine those cases where $B$ is true and then for each of those, find the proportion where A is also true. This will yield the same result as method 1.

> ##Method 2
> #For the reduced sample space where B is True, What proportion have A=True?
> P_of_A_given_B_m2 <- sum(mydata$A[mydata$B==1])/sum(mydata$B==1)
> P_of_A_given_B_m2
[1] 0.2807018

So, in this example, $P(A|B)=0.2807018$. Of course, there are several ways to go about getting these more quickly using various functions in R, but this shows the computation in a way that's hopefully easier for you to understand the probability concepts as well as the computation in R.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.