How do I use the "axis" parameter in pandas?

Key Takeaways

hello and welcome back to my Q&A video series about the panda library in Python and the question for today is how do I use the axis parameter in pandas all right this is a bit of a technical topic uh we've seen the axis parameter once before in this series in the context of dropping rows and columns so we're going to start uh with a recap of that we're going to talk about the AIS parameter and then we're going to use it in another context okay so uh as always we'll start by importing pandas import pandas as PD and then we're going to read in a data set and this is going to be a data set of alcohol consumption by country so pd. read CSV and then pass it the URL bit. Le SL drinks by my country okay so we'll run that and then take a look at the first five rows and what we see is that each row represents a country and uh their reported alcohol consumption per adult okay so uh let's say we want to remove a column let's say we don't care about the continent column we would what we would do is use the uh drop method it's a data frame method so drinks. drop and then we pass it as a string the name of the column we want to drop and then we say axis equals one okay and uh I'm just going to go ahead and throw on the head method so that we're only looking at the first five rows of the resulting data frame now what you can see is that the continent column was indeed dropped okay and I had to specify axis equals 1 to say I want to drop a column okay now uh this did not I did not use the in place parameter um so this did not actually remove the column this is just a temporary thing okay so let's say instead I wanted to drop a row so I wanted to drop row two for example this is the index and uh I'll so instead I say drop two and I change it to axis equals z and this time you can see the continent uh column is back but this time row two is now gone okay so uh what exactly is going on here so uh pandas could have created a drop columns method that was specific to drop dropping columns and it could have created a separate drop rows method for dropping rows but instead the panda philosophy is to make a single method called drop and then we use the axis parameter to specify which what we want to drop do I want to drop a row or a column so axis one is the column axis axis zero is the row axis okay so uh anytime you run across a pandas command uh and you think it could refer to rows or columns that's how you'll specify which you want to refer to okay now let's try using axis in a different parameter um so what we're going to do is a mathematical operation a mean and let's just say I say drinks. mean what's going to happen so I'll run that and what we're seeing is that pandas has told us the mean of each of the numeric columns this is the mean value of the beer servings series of the spirit serving series wine servings and total leaders of pure alcohol okay so why did it give us the mean values for each series um great question so basically this is doing column means and that's because the default behavior of mean is Axis equals z so I'm just making it explicit this will have the same result I'm just making it explicit that when I said drinks. mean it was assuming the zero axis and in other words axis zero for an operation is saying I want you to move down I want you to move across the row axis this is the row axis and the direction I want the operation to occur is down so the way I like to imagine it is that these four columns the numeric columns are kind of being collapsed down into a single set of four numbers that represents the mean of each column okay so if instead you wanted the mean of each row you could say axis equals 1 and now you can picture that operation moving the mean operation moving along the column axis the one one axis it's moving this way and that way what it's doing is it's averaging it's taking the mean of these zeros which is zero it's taking the mean of these which is 69 etc etc now this is not a uh meaningful number because these are different units these are servings this is leaders so they're different units so this isn't like a meaningful number but um I just wanted to show you the principle of how to use the a of0 or one okay so uh just to reinforce this when you're moving along axis one you're preserving all the rows and there's 193 rows which is why uh that result has a shape of 193 whereas when we did axis equals z we got a shape of four because it was collapsing all the columns down into one number so again I think of the axis for an OP a mathematical operation as like the direction of the movement of the operation okay all right so as always we're going to end with a bonus tip and one one thing I want to show you is that you can actually they're actually aliases for the axis numbers so if you find zero you might find this more intuitive than axis Z and one um so if you want to say axis zero you can actually say axis equals index as a string and you'll get the same thing as axis equals z okay axis equals 1 which again would have given us 193 numbers this way you can say axis equals columns okay so you decide which is more intuitive um up to you I tend to use zero and one um but index and columns is also valid in pandas all right so that's it for today thanks so much for joining me uh feel free to click subscribe if you'd like to see more videos like this if you have a pandas tip or a question please let me know in the comment section below so that we can all help each other learn and um I hope to see you again soon thanks again