1 00:00:02,230 --> 00:00:04,450 Everyone and welcome back to this class 2 00:00:07,800 --> 00:00:13,290 in this lecture I'm going to go over a better way to install data science and machine learning libraries 3 00:00:13,290 --> 00:00:16,110 for Python for Windows users. 4 00:00:16,140 --> 00:00:20,820 Historically Windows users have had a lot of problems installing this stuff. 5 00:00:20,820 --> 00:00:26,610 Luckily these days there is an option that makes things very painless and just as easy as they are on 6 00:00:26,610 --> 00:00:30,090 Linux or Mac that is Anaconda. 7 00:00:30,090 --> 00:00:34,280 In fact even if you're not on Windows you can still use Anaconda. 8 00:00:34,290 --> 00:00:38,940 It's nice because it isolates your environment from the defaults provided on your system. 9 00:00:38,940 --> 00:00:45,840 So for example you can have Python 3 in Anaconda but Python to as your system default. 10 00:00:45,890 --> 00:00:50,990 When I first started these courses I wasn't keen on Windows since there were a few essential libraries 11 00:00:50,990 --> 00:00:54,820 that couldn't be installed on Windows without a significant amount of effort. 12 00:00:54,920 --> 00:01:01,470 If at all in my view anything beyond a couple of lines in the console or clicking an install file is 13 00:01:01,470 --> 00:01:02,300 too much. 14 00:01:02,460 --> 00:01:07,510 And believe me some students even have trouble with that so it's good not to make things too complicated 15 00:01:07,530 --> 00:01:09,650 before you can even begin the course. 16 00:01:10,180 --> 00:01:12,150 Nowadays that has changed. 17 00:01:12,480 --> 00:01:18,450 It's a lot easier to install things on Windows in large part thanks to Anaconda and so this lecture 18 00:01:18,450 --> 00:01:23,460 is all about how to install all the data science and machine learning libraries you'll need on Windows 19 00:01:23,550 --> 00:01:25,650 using Anaconda. 20 00:01:25,650 --> 00:01:30,900 So in this lecture I'm going to walk you through how to install Anaconda as well as some of the libraries 21 00:01:30,900 --> 00:01:34,080 you might need that don't already come with Anaconda. 22 00:01:34,620 --> 00:01:39,580 You'll find that most of the common libraries such as pie and Type-I are already included. 23 00:01:39,600 --> 00:01:43,340 So if that's all you want to use then for you it's just a one click install 24 00:01:48,360 --> 00:01:49,110 on this slide. 25 00:01:49,110 --> 00:01:53,730 I'm going to give you a super short summerize version of this lecture so you don't have to walk through 26 00:01:53,730 --> 00:01:56,440 the installation with me if you don't want to. 27 00:01:56,490 --> 00:01:59,110 For some people that really helps since you can see it. 28 00:01:59,120 --> 00:02:04,790 But if you can do it on your own feel free so number one. 29 00:02:04,930 --> 00:02:07,270 Download and install Anaconda. 30 00:02:07,270 --> 00:02:09,039 This is just a one click install. 31 00:02:09,070 --> 00:02:13,290 It already includes some pie Type-I that plug lib and panderers. 32 00:02:13,480 --> 00:02:15,820 That's all we need for the stack in Python. 33 00:02:15,850 --> 00:02:20,890 Linear regression and logistic regression and a few more courses. 34 00:02:20,930 --> 00:02:26,630 It also comes with T.K. which is what we use for an IP and psyche learn which has some pre-built machine 35 00:02:26,630 --> 00:02:29,660 learning models. 36 00:02:29,920 --> 00:02:34,980 Now even though this stuff comes by default you can still update them if you want so you can do Konda 37 00:02:34,990 --> 00:02:43,530 update an umpire for example and that will update by number to install deep learning libraries. 38 00:02:43,540 --> 00:02:49,270 We've got Pipp installed tensor flow that's going to install tens of flow and if you want to install 39 00:02:49,270 --> 00:02:55,590 keras you have to first do Conda in stoppit which is going to update Pipp and then you can install Karris 40 00:02:55,620 --> 00:02:59,620 using pipin Stolle care us if you don't update pitte first. 41 00:02:59,620 --> 00:03:05,790 You might get an error X we have a c and T.K. which is Microsoft's deepening library. 42 00:03:05,840 --> 00:03:11,870 So you do Pipp install and then the C and Teekay euro which you can get from Microsoft's Web site. 43 00:03:12,110 --> 00:03:17,170 So I'm not posting any you are all here because the version could likely change in the future. 44 00:03:17,170 --> 00:03:22,430 And so you can Google search how to install CNC K and you can get a euro just like this. 45 00:03:24,550 --> 00:03:31,560 Next we have pie torche that's Konda and start minus see Peter J.C 2:59 pie torche. 46 00:03:31,580 --> 00:03:38,270 After that we have CNO So that's Konda and start CNO or Kanda install Viendo pied you. 47 00:03:38,420 --> 00:03:42,670 If you have an Nvidia you and you've installed the khud a tool kit already 48 00:03:46,210 --> 00:03:48,790 number three install open aiight Jim. 49 00:03:48,790 --> 00:03:54,550 That's just Pippin's start Jim if you want to be able to play Atari games also then that's more involved 50 00:03:54,550 --> 00:04:01,030 so just skip to the end of this lecture where I walk through that if you want to play and save videos 51 00:04:01,030 --> 00:04:10,210 using open gym then you also have to install FFM pick. 52 00:04:10,290 --> 00:04:13,760 So the first thing we're going to do is go over to the Anaconda Web site. 53 00:04:13,830 --> 00:04:16,980 That's an icon a dot com slash download. 54 00:04:17,040 --> 00:04:23,290 Scroll down to the window section and click on either Python 3.6 or Python 2.7. 55 00:04:23,490 --> 00:04:26,410 Or if you're watching this lecture in the future and there's a new version. 56 00:04:26,430 --> 00:04:32,180 Get that the code in my courses is compatible with both Python 2 and Python 3. 57 00:04:32,190 --> 00:04:37,710 So in that regard it doesn't really matter which when you get in the lectures you might see Python to 58 00:04:37,740 --> 00:04:38,180 code. 59 00:04:38,280 --> 00:04:43,470 But the best way to make sure you're seeing the latest version is to get Paul inside the course repo 60 00:04:44,860 --> 00:04:49,430 so make sure you're always doing that because I'm constantly making new updates. 61 00:04:49,430 --> 00:04:54,070 Now though Python 3 is newer There are still reasons to use Python too. 62 00:04:54,170 --> 00:05:00,920 For example in your work you might use Python 2 or certain platforms like Google App Engine only support 63 00:05:01,130 --> 00:05:02,370 Python too. 64 00:05:02,540 --> 00:05:06,120 So if you are running a web app that means you're stuck with Python too. 65 00:05:06,170 --> 00:05:12,260 It does have great scalability features so there are many good reasons to use Google App Engine if you 66 00:05:12,260 --> 00:05:16,420 want to get more insight on whether to choose Python to Python 3. 67 00:05:16,430 --> 00:05:24,030 Just check out the appendix lecture Python 2 vs Python 3. 68 00:05:24,090 --> 00:05:28,550 So now that we've downloaded the install file all we need to do is click on it. 69 00:05:28,590 --> 00:05:43,070 That's what I mean by one click install you click on this here OK a few times and everything is done. 70 00:05:43,090 --> 00:05:47,620 Unfortunately I gave my username a space which kind of sucks but that's what happened. 71 00:05:47,890 --> 00:05:50,570 So I'm sure some of you have a space in your username too. 72 00:05:50,590 --> 00:05:54,180 So if I come across any issues at least you'll know what to do. 73 00:06:04,070 --> 00:06:05,880 Alright so everything's installed. 74 00:06:05,960 --> 00:06:12,320 So essentially everything except the differing libraries have already been automatically installed so 75 00:06:12,320 --> 00:06:19,960 you don't need to manually install them pie Sipar map like live Pandurs Python or sikat learn. 76 00:06:20,200 --> 00:06:25,690 So if you're taking my numpad course or any course that doesn't use modern deep learning libraries you 77 00:06:25,690 --> 00:06:27,850 already have everything you need. 78 00:06:28,060 --> 00:06:31,030 So let's go into Python and make sure that's the case. 79 00:06:32,510 --> 00:06:39,350 To start Python I go to the start menu type in Anaconda and then this anaconda prompt application should 80 00:06:39,350 --> 00:06:40,360 pop up. 81 00:06:40,370 --> 00:06:51,860 Actually it should pop up before you even finish typing Anaconda. 82 00:06:51,920 --> 00:06:56,450 So we go into there and this brings up a command line terminal next. 83 00:06:56,480 --> 00:07:05,410 Just type in Python. 84 00:07:05,430 --> 00:07:08,530 After that we can import all the libraries I mentioned earlier. 85 00:07:08,610 --> 00:07:40,600 If we don't get an error that means they've been installed successfully. 86 00:07:40,730 --> 00:07:44,800 So let's try something simple like generating some random numbers and making a plot 87 00:08:00,580 --> 00:08:02,540 so that's a plot of random noise. 88 00:08:06,220 --> 00:08:07,740 Let's make a histogram to 89 00:08:18,480 --> 00:08:21,410 it so we see a normal curve just like we expect. 90 00:08:27,160 --> 00:08:31,370 Now you can see that tensor flow is not installed which is why we get this error. 91 00:08:31,650 --> 00:08:38,150 But we can install it very easily by exiting a python and then typing in Pipp install tensor flow 92 00:08:47,090 --> 00:08:47,780 next. 93 00:08:47,780 --> 00:08:49,370 Let's try to install Cara's 94 00:08:52,330 --> 00:08:53,760 so we get this error. 95 00:08:53,800 --> 00:09:00,540 So I looked this up and determine that we need to update Pitt so let's do that by typing in Konda install 96 00:09:00,550 --> 00:09:04,070 Pipp Kondo kind of works like pipin that way. 97 00:09:04,080 --> 00:09:06,300 They are all just tools for installing stuff. 98 00:09:11,420 --> 00:09:22,190 Now let's try hip install Karris again. 99 00:09:22,190 --> 00:09:22,530 All right. 100 00:09:22,560 --> 00:09:23,530 So everything works 101 00:09:26,810 --> 00:09:31,100 next let's try to install and T.K. this is used in my and AP courses 102 00:09:36,140 --> 00:09:39,440 so it looks like it's already installed so there's nothing more to do. 103 00:09:40,830 --> 00:09:43,910 Just keep in mind if we come across a library you don't care about. 104 00:09:43,920 --> 00:09:45,530 Feel free to ignore it. 105 00:09:45,600 --> 00:09:50,220 I find it's useful just to install everything at the same time so that when you're deep in the code 106 00:09:50,220 --> 00:09:52,920 later you don't have to think about stuff like this. 107 00:09:56,160 --> 00:09:59,100 So the next thing we'll install is see A.K.. 108 00:09:59,400 --> 00:10:02,540 This is Microsoft's deep learning library. 109 00:10:02,550 --> 00:10:04,270 Notice how it's not part of Pipp. 110 00:10:04,320 --> 00:10:10,480 So you need to Gravier your role manually from Microsoft's Web site. 111 00:10:10,490 --> 00:10:14,870 Unfortunately it's a little hard to find because there are many pages that deal with how to install 112 00:10:15,010 --> 00:10:17,490 s.a.a on Microsoft's Web site. 113 00:10:17,780 --> 00:10:22,180 But what you're looking for is a page that has a link to a W H L file. 114 00:10:22,460 --> 00:10:27,950 So copy and paste that after Pipp install this is a good example because it shows you another way you 115 00:10:27,950 --> 00:10:31,180 can use Pipp by doing Pipp install and then a URL 116 00:10:46,750 --> 00:10:47,270 next. 117 00:10:47,290 --> 00:10:49,510 Let's install PI torch. 118 00:10:49,510 --> 00:10:56,730 This requires us to use a custom source so we have to specify the option minus see Peter J.C 1 2 three. 119 00:10:57,040 --> 00:11:02,460 So that's Konda install minus see Peter J.C 2:59 PI torch. 120 00:11:02,470 --> 00:11:09,130 This is because a very nice guy called PIERGROSSI 2:59 has provided us with a version of Pi torch that 121 00:11:09,130 --> 00:11:12,350 works on Windows. 122 00:11:12,540 --> 00:11:18,070 And once that's all done we can verify that they've been installed correctly by going back to Python. 123 00:11:33,240 --> 00:11:33,530 All right. 124 00:11:33,530 --> 00:11:35,250 So tensor flow works. 125 00:11:37,120 --> 00:11:38,070 Cara's works 126 00:11:41,490 --> 00:11:42,660 in Teekay works 127 00:11:45,760 --> 00:11:47,030 s.a.a works 128 00:11:50,440 --> 00:11:51,430 and torche works 129 00:11:59,350 --> 00:12:06,130 next lesson stuff yanno Viendo has historically been pretty difficult to install on Windows but nowadays 130 00:12:06,430 --> 00:12:07,950 that's not the case. 131 00:12:08,020 --> 00:12:11,270 So if you go to their Web site you'll see a bunch of instructions. 132 00:12:11,410 --> 00:12:17,230 If you don't want to use the GPS you or you don't have a GP you then the instructions will be very easy. 133 00:12:17,470 --> 00:12:22,400 I don't have keep you on this machine so I'm going to do the easy version. 134 00:12:22,420 --> 00:12:25,490 Keep in mind that Viendo is really great for learning purposes. 135 00:12:25,510 --> 00:12:26,860 So it's totally fine. 136 00:12:26,860 --> 00:12:32,980 Even if you have a AGP you to just install the CPQ version for now and then use the GPU version of tensor 137 00:12:32,990 --> 00:12:34,900 for now. 138 00:12:34,900 --> 00:12:40,010 I ended up upgrading Cale's service in lib Python since that's what they told me to do on Vienna's Web 139 00:12:40,020 --> 00:12:40,710 site. 140 00:12:40,840 --> 00:12:43,030 But it looks like these were already installed. 141 00:12:43,030 --> 00:12:46,440 In fact updating MKR service gave me an issue later. 142 00:12:46,450 --> 00:12:47,740 So we'll have to fix that. 143 00:12:49,280 --> 00:12:55,680 So if you want to install CNO for you only That's kinda install Viendo if you want to install Viendo 144 00:12:55,700 --> 00:13:29,380 for the cheap you then do Conder install Jano you. 145 00:13:29,420 --> 00:13:32,680 Now let me go into Python and check if CNO works. 146 00:13:41,930 --> 00:13:46,860 So we get an error because of this MKR service thing that I mentioned earlier. 147 00:13:46,970 --> 00:13:48,970 So let's set this environment variable. 148 00:13:48,980 --> 00:13:58,160 It's telling me to set. 149 00:13:58,320 --> 00:14:03,780 By the way this is great to know if you don't yet know how to set environment variables on Windows. 150 00:14:03,930 --> 00:14:14,130 We can also check that it works by using the echo command. 151 00:14:14,170 --> 00:14:15,510 So let's try that again. 152 00:14:16,770 --> 00:14:19,700 Let's do a simple example of adding two numbers MVNO. 153 00:14:19,740 --> 00:14:21,540 Just to make sure everything's working. 154 00:14:59,160 --> 00:15:04,170 If you want to do something even more complicated you can run this script from replaying part 2 which 155 00:15:04,170 --> 00:15:06,730 doesn't require any external data set. 156 00:15:06,780 --> 00:15:12,100 So just go over to the folder and in class to type in Python grid search. 157 00:15:12,130 --> 00:15:13,430 Hi. 158 00:15:13,480 --> 00:15:23,200 So that's going to look for hyper parameters using cross-validation. 159 00:15:23,270 --> 00:15:28,750 Now in this last section of this lecture we're going to talk about reinforcement learning when we started 160 00:15:28,750 --> 00:15:30,340 studying reinforcement learning. 161 00:15:30,350 --> 00:15:34,880 There is yet another library will need to install called Open a gem. 162 00:15:35,120 --> 00:15:39,310 If you don't plan on learning reinforcement learning you can skip this part of the lecture. 163 00:15:40,140 --> 00:15:45,810 This has also historically been very difficult but luckily the open source community has put in the 164 00:15:45,810 --> 00:15:46,200 work. 165 00:15:46,230 --> 00:15:47,910 So you don't have to. 166 00:15:47,910 --> 00:15:50,120 You're welcome to read through the get her issue on this. 167 00:15:50,130 --> 00:15:54,210 If you want but I'm going to just do the simplest thing that works. 168 00:15:54,480 --> 00:15:56,580 So lets first do Pippin's Stajan 169 00:16:05,820 --> 00:16:10,560 great so that worked. 170 00:16:10,580 --> 00:16:12,560 Now the second command is a bit longer. 171 00:16:12,590 --> 00:16:19,010 So let's go to the actual get hub issue and copy and paste the command the easiest way to get there 172 00:16:19,010 --> 00:16:25,020 is just to go to Google and type in install open gym windows anaconda or something along those lines. 173 00:16:49,680 --> 00:16:51,130 So let's paste that in. 174 00:17:00,290 --> 00:17:03,410 And notice how I feel since I haven't yet installed it. 175 00:17:03,650 --> 00:17:10,780 So we can install it by doing Konda install get. 176 00:17:10,780 --> 00:17:12,220 Now let's try it again. 177 00:17:18,569 --> 00:17:23,670 Unfortunately this fails again because we need gcc which is a C compiler. 178 00:17:23,670 --> 00:17:29,440 Now one way to get gcc is to do Konda install MTU w 64 toolchain. 179 00:17:29,580 --> 00:17:32,510 But unfortunately I tried this and it also doesn't work. 180 00:17:32,520 --> 00:17:35,340 In fact I think this toolchain was installed already. 181 00:17:35,670 --> 00:17:37,950 So I tried quite a few things that didn't work. 182 00:17:37,950 --> 00:17:43,590 So in order to save you some time I'm going to recommend you only try this stuff on this page. 183 00:17:43,710 --> 00:17:45,710 If everything else doesn't work for you. 184 00:17:46,110 --> 00:17:51,330 So it worked for me was to just grab the pre-compiled binary directly. 185 00:17:51,520 --> 00:17:56,260 Now in order to do that you want to go to Jolies get help repo directly. 186 00:17:56,330 --> 00:17:58,170 So that's your role here. 187 00:17:58,430 --> 00:18:05,820 Get up dot slash cordiale slash Atari dash pie slash releases. 188 00:18:05,970 --> 00:18:10,600 Next you'll want to download the W h file that matches your environment. 189 00:18:10,780 --> 00:18:15,440 So I have Python 3.6 on a 64 bit installation of Windows. 190 00:18:15,520 --> 00:18:18,180 So this is a file I want. 191 00:18:18,280 --> 00:18:22,700 Luckily we already discussed earlier in this lecture how to install a w h file. 192 00:18:22,750 --> 00:18:23,920 That's just Pipp. 193 00:18:23,950 --> 00:18:25,420 And then the path to the file. 194 00:18:25,420 --> 00:18:40,710 So let's do that. 195 00:18:40,880 --> 00:18:45,830 Now we can test our installation by running a script that requires an Atari game. 196 00:18:45,830 --> 00:18:49,510 So let's see over to RL too and then Atari. 197 00:18:49,520 --> 00:18:52,140 And now let's run Dijk UN underscore TFT. 198 00:18:52,180 --> 00:18:52,520 Hi 199 00:19:00,150 --> 00:19:01,600 cool so everything's good. 200 00:19:08,460 --> 00:19:15,940 Now the final thing we want to do for open gym is if you want to play a video or save a video you want 201 00:19:15,940 --> 00:19:29,810 to install F-F MPEG so to do that you want to type in Konda install mynahs see men mento F-F MPEG. 202 00:19:29,830 --> 00:19:33,880 Once you've done that you can go to the card pool folder and type in Python. 203 00:19:33,880 --> 00:19:41,680 Save a video PI and this will run a script that will play the card pull game show a plot and then save 204 00:19:41,680 --> 00:19:42,400 a video 205 00:19:49,430 --> 00:19:54,210 and of course you can also play this video by just navigating to the file and clicking on it. 206 00:20:08,760 --> 00:20:10,370 So for now that's everything. 207 00:20:10,500 --> 00:20:15,800 If I need to add new libraries or updates to this lecture they will be appended at the end.