1 00:00:05,200 --> 00:00:09,360 For this example, we're going to select candidates to interview for a job. 2 00:00:09,360 --> 00:00:13,120 We've identified a list of core skills that successful candidates must have, 3 00:00:13,120 --> 00:00:15,840 in order to perform the role we're hiring for. 4 00:00:15,840 --> 00:00:18,960 I've created a new Python file called candidates.py, 5 00:00:18,960 --> 00:00:25,668 for this example. Pause the video while you create your file, then type the required_skills list. 6 00:00:40,000 --> 00:00:44,000 I've kept the list short, to make it easier to verify our results. 7 00:00:44,000 --> 00:00:47,120 The HR department have processed all the job applications, 8 00:00:47,120 --> 00:00:51,397 and their system has produced a dictionary containing the details for the candidates: 9 00:03:02,720 --> 00:03:07,268 We need to check the skills for each candidate, to decide whether or not to interview them. 10 00:03:07,268 --> 00:03:12,480 We'll create an empty set, to hold the names of the interviewees, then check their skills. 11 00:03:12,480 --> 00:03:16,526 I'll type the code first, then explain how we perform that check: 12 00:03:50,480 --> 00:03:54,160 Ok, how does this work ? We iterate over the candidates 13 00:03:54,160 --> 00:03:58,240 in the dictionary, to get each candidate name, and their skills. 14 00:03:58,240 --> 00:04:03,440 As you can see on lines 4 to 9, the skills are stored as sets. This approach would still work 15 00:04:03,440 --> 00:04:08,720 if they were stored in lists or tuples - we'd just use the set function to convert them first. 16 00:04:08,720 --> 00:04:14,480 Or, we could convert our required_skills to a set, and perform the check the other way round. 17 00:04:14,480 --> 00:04:19,040 Next, we want to check that the candidates' skills contain all the skills we require. 18 00:04:19,040 --> 00:04:23,920 If a set contains all the items from another set, then it's a superset of the other set. 19 00:04:23,920 --> 00:04:27,760 The check is very simple - we just use the issuperset method. 20 00:04:27,760 --> 00:04:31,920 If it returns True, we add the candidate to our set of interviewees. 21 00:04:31,920 --> 00:04:36,480 That's checked on line 14. If you try to do this without using sets, you'll 22 00:04:36,480 --> 00:04:42,640 end up writing nested loops, and a lot more code. Sets are very powerful for this sort of problem. 23 00:04:42,640 --> 00:04:46,480 Ok, does it work ? Looking at our candidates' skills, 24 00:04:46,480 --> 00:04:51,452 Anna has everything we need. Bob also has the 3 skills we're looking for. 25 00:04:51,452 --> 00:04:55,840 Daniel doesn't - he hasn't listed Python or Linux amongst his skills. 26 00:04:55,840 --> 00:05:00,960 Ekani also has what we're looking for. Fenna doesn't have Python – which is a shame, 27 00:05:00,960 --> 00:05:04,960 because she has some impressive skills there. Python has borrowed aspects of 28 00:05:04,960 --> 00:05:09,200 all the languages she's listed, so a human might well decide to interview her. 29 00:05:09,200 --> 00:05:12,748 But I digress, let's run the program and see what we get. 30 00:05:14,800 --> 00:05:20,080 We get the four candidates that we expect: Anna, Bob, Carol, and Ekani. 31 00:05:20,080 --> 00:05:25,120 If there were a lot of candidates, this could be a way to select a small enough set to interview. 32 00:05:25,120 --> 00:05:28,240 You could also use this approach in an on-line dating app. 33 00:05:28,240 --> 00:05:33,440 Tinder uses the Python language for some of its back-end server code. Of course, I can't say 34 00:05:33,440 --> 00:05:38,625 that they use Python sets to group potential matches, but you can see how that could work. 35 00:05:38,625 --> 00:05:43,569 I'll finish this video with an example of the difference between a subset, and a proper subset. 36 00:05:43,569 --> 00:05:49,037 This data gives us an opportunity to see the difference in practice. What a coincidence! 37 00:05:49,037 --> 00:05:54,400 We're very short of time, and can't spare people to interview many candidates. We need to reduce 38 00:05:54,400 --> 00:05:58,560 the number of candidates we've selected. At the moment, we're matching anyone who 39 00:05:58,560 --> 00:06:04,480 has the skills we require. That includes Bob, who has what we need, but nothing more. 40 00:06:04,480 --> 00:06:10,874 If I had to remove someone from our selection, Bob would be my choice. Let's see if Python agrees. 41 00:06:10,874 --> 00:06:14,240 We checked if the candidates' skills are a superset of our required skills, 42 00:06:14,240 --> 00:06:18,800 which means we also get people who's skill set is equal to our required skills. 43 00:06:18,800 --> 00:06:23,040 If we check for a proper superset, instead, we should end up with people who have more than 44 00:06:23,040 --> 00:06:28,400 what we require. That's not a bad criterion for reducing the list, so let's do it. 45 00:06:28,400 --> 00:06:32,520 I'll change the test on line 14. I won't delete that line - 46 00:06:32,520 --> 00:06:37,256 I'll comment it out and add the new line below, so that you can compare the two bits of code: 47 00:06:48,296 --> 00:06:52,800 required_skills is a list. I did that as a reminder that you can pass any 48 00:06:52,800 --> 00:06:57,600 iterable to these set methods. You can see that in the commented out line 14. 49 00:06:57,600 --> 00:07:02,240 So on line 15, I've had to convert required_skills to a set. 50 00:07:02,240 --> 00:07:06,000 That's not very efficient, converting it each time round the loop. 51 00:07:06,000 --> 00:07:09,680 In practice, you'd either convert it once, before the loop starts, 52 00:07:09,680 --> 00:07:14,000 or create it as a set on line 1. I don't want to change the code too much, 53 00:07:14,000 --> 00:07:18,563 on video, so I'll leave it like that. Just be aware of the performance implications of 54 00:07:18,563 --> 00:07:24,560 performing an operation like that, inside a loop, when you could do it once before the loop starts. 55 00:07:24,560 --> 00:07:28,400 Alright, does it work ? Run the program: 56 00:07:29,800 --> 00:07:33,760 and we've eliminated Bob. We used the proper superset operator, 57 00:07:33,760 --> 00:07:38,880 greater than, on line 15. That only returns True if the skills are a proper superset of our 58 00:07:38,880 --> 00:07:44,640 required skills – which means they must include at least one more skill than the required_skills set. 59 00:07:44,640 --> 00:07:48,443 That demonstrates the difference between a subset (or superset) 60 00:07:48,443 --> 00:07:53,680 and a proper subset (or superset), and gives an example of how they change the results. 61 00:07:53,680 --> 00:07:56,720 And that's the end of our discussion on Python sets. 62 00:07:56,720 --> 00:08:01,280 The next lecture is a document - a summary of everything we've covered in this section. 63 00:08:01,280 --> 00:08:05,900 Make sure you're comfortable with all the topics covered, and I'll see you in the next section.