When we're talking about digitizing and data creation, whether you're digitizing the data or you're getting it from somewhere else, it's a nice place to stick in a little conversation about accuracy versus precision. Often those terms get used by people interchangeably, but they actually mean different things. It's important to understand what they mean in relation to either creating your own data, and or getting data from somewhere else, and being able to assess what is the accuracy or precision of that data. Okay. Here we have an aerial image of the University of Toronto campus. If we zoom in here, you'll notice that this is what we call the front campus at this location. That's known as the front campus. It's just a field. So, if we can continue to zoom in here, there's actually a fire hydrant located at that point there. It's actually one of the more scenic fire hydrants on campus. It's got this beautiful building in the background. That's beside the point though. All right. So, I actually went out to this fire hydrants. If you were trying to collect data for this point, you wanted to be able to show it on your map, then you'd want to try and get it as accurate as possible in terms of the position of that fire hydrant. So, this is the true location of the fire hydrant. This is where it actually exists in the real world. You'd have to say to yourself, "Well, how do I want to represent that fire hydrant, is a teeny tiny little point, is an area, what am I going to do?" So, this is really as much specificity as you'd be able to get in terms of being able to show that. There's no point in creating it as anything smaller than that because this is the actual size of the physical object there. So, this is our true location of this hydrant. So, what I did was I took my phone. This is actually an older version of my phone that I had before, but hey, it still works, and I have this great app that I was using for it called MotionX-GPX. I was able to put my phone right on top of the fire hydrant and collect the longitude and latitude for that location for that fire hydrants. So, you'll see here that I then switched it to universal transverse Mercator coordinates, which are in meters, which I find much easier to work with. You'll see that we have our Eastings and Northings. So, these are the coordinates in meters. At that time, when I first started trying to collect the coordinates for this, it was giving me an accuracy of 47 meters. So, I'll show you what that means in a second. I waited around and the longer I waited, so you'll notice actually it wasn't that long, the timestamp here is 10:05, and this was only a minute later, that we went from an accuracy of 47 meters to an accuracy of 17 meters, which is a lot better. Not amazing, but better. Certainly not the size of a fire hydrant, but that's about as good as I was able to get there in terms of accuracy. So, what does this actually really mean? Well, if we zoom out a bit here, if I had an accuracy of 17 meters, then what that really means is all I know when I put that point on the map is that it's somewhere inside that 17 meter circle. It could be here, it could be here, could be over here, could be anywhere. I don't really know if you do them as plus signs or crosses, whatever, is that it's somewhere inside that circle. So, that's an accuracy of 17 meters. But, you'll notice that the precision that I was using is one meter. Why is that one meter? Because if we go back here for a second, there's no decimal place here for this coordinate. It's just telling me 629329 meters East. That's it. There's no point 235 or whatever. So, that means that the precision of that is 21 meter. So, that's why when I have it up here, I'm saying it has a precision of one meter and it has an accuracy of 17 meters. If I had an accuracy of the original 47 meters, then that's actually much worse. You can see that all I know is that it's somewhere inside that much larger circle. Okay? So, let's talk about accuracy and precision. What are the definitions of them and what are we actually talking about here. So, for this fire hydrants, accuracy can be defined as the difference between the recorded value and the true value. So, the true value is where the thing actually really is in reality. The recorded value was where I was able to say that that thing was in my dataset. A common analogy that's used for accuracy and precision is using a dartboard. I like it. This is actually from an old textbook that I used to use by Lowe and Young, but I quite like it, is still that. If you're throwing darts at a dartboard, and you hit the bull's eye every time, and that's what you're aiming for, then that would be very accurate. You're able to throw the darts very accurately. The analogy here is that when you're collecting points for fire hydrants or for that particular fire hydrant, if you were able to collect those points at the actual location, that would be very accurate. Now, precision is the fineness of the measurements. Not the accuracy. That's a different thing. So, I like this one is that I've got this clock here, and if I ask you what time it is, and you tell me that it's February. Let's say that it really was at that time, then that's accurate, but it's not very precise. So, it's not a very fine measurement. If it actually is a February, then it's like well, yes it is February, but could you help me out a little bit here? Could you get a little better? So, if I ask you again and you said Wednesday, that's more precise. It's still accurate, but we have a higher level of precision. So, now instead of it being down to the month, it's down to the day. If I ask you again and you said morning, that's even more precise. If you said 5:23 a.m., that would be even more precise, which is actually what time it's indicating on the clock there. So, less precision, more precision, more precision, as we go. So, that's the difference. So, all of these are accurate. But, we can change the amount of precision that we have to increase, I would assume, the utility or the usefulness of the data that we're able to collect. Going back to our dartboard analogy, is that if we are able to collect points that are all close together to one another, we can have a high level of precision, but have a low level of accuracy. So, I think a nice example of this is you can have a watch, and if you're collecting times of say when people come into work in the morning, when they go through the door to the office, and you could be collecting those down to the second, and saying okay Bob came in at 10:43 and 23 seconds. So, that's very precise. When you might record all these and then realize later that your watch was actually 10 minutes fast. So, that's not very accurate because all of your measurements are off by 10 minutes. But, it was precise because you got them down to the second. So, that's what they're trying to show here. What I'm trying to show with this dartboard is that your measurements are all close together. The fineness of the measurement is there, say down to the second, but it's off from the accurate measurement of reality of what you were trying to actually measure. So, that's the fineness of the measurement. So, in terms of precision and accuracy in something like a GIS like a ArcMap, is often by default it will give you a number, if you measure a distance, it'll give you a number to six decimal places. Well, that's actually ridiculous. I mean we're actually talking here about a millionth of a meter and it's very rare that you're going to get measurements that are actually that precise in reality. So, the software in other words is giving you this false sense of precision that's not really there. So, just because it's giving you those six decimal places and making you think that you're measuring things down to a micron, it's not really. That's a high level of precision, but it's not a high level of accuracy. Sometimes people will be loathe in the sense of thinking well we'll look at all those decimal places, it must be really accurate then. No. Not accurate. Just precise. So, just to finish that off is that if the actual distance is 50 meters, then we have something that's very precise, but not very accurate. To finish off our dartboard example, so we can have a high precision, but low accuracy. We can have a low precision, but a high accuracy. So, in other words, these are all close to being the real value, but they're not very precise. We can have a level of high precision and high accuracy, which is probably the most desirable, or we can have the worst case scenario, which is low precision and low accuracy. If we go back to our dartboard example, then maybe you've had a few too many pitchers of beer or something, and so you've lost your precision and your accuracy. So, that's not exactly ideal. So, I hope that makes sense in terms of what precision and accuracy means. To relate this back to our digitizing and our data model, let's have a look at the map again. If you're creating new features and digitizing them, if we zoom in here, you can see that this is precise. We might be digitizing this down to say within a meter, but it's not accurate because it's not actually in the right place, the lines here. So, you can see there, this is easier to see with the yellow. So, this is precise, but not accurate. So, it's important to think about this when you're getting data, at what scale was it digitized? is it appropriate for my purpose when you're creating your own data or getting data from somebody else?