WEBVTT 00:01.270 --> 00:07.960 All right, so last step, we know a restaurant review classification project is applying some machine 00:07.960 --> 00:09.670 learning classification algorithm. 00:10.240 --> 00:14.760 Now, before that, we have to split up a site and forte as well. 00:14.980 --> 00:19.670 You know, earlier projects are sort of really like a train split. 00:20.640 --> 00:23.140 So from model selection model of this scale. 00:23.140 --> 00:28.270 And we are going to split this into X, train X, this white linen like this. 00:28.690 --> 00:31.210 And I just kept this size 20, wasn't it? 00:31.300 --> 00:34.260 So out of thousand, record twenty was in the job record. 00:34.270 --> 00:38.950 We'll go to testing buckets and remaining 80 percent leading to raining buckets. 00:39.450 --> 00:43.120 Let's just keep a random street detail just to recreate the same result. 00:43.450 --> 00:44.040 The Orango. 00:44.040 --> 00:44.240 So. 00:45.190 --> 00:45.580 All right. 00:46.510 --> 00:51.400 Let me display the shape of this X and this train. 00:54.750 --> 00:57.760 And extends this dog --. 00:59.790 --> 01:01.130 Offsets a small little. 01:03.440 --> 01:10.640 So there will be a 800 records assigned to cleaning buckets and 200 records assigned to testing buckets. 01:11.210 --> 01:21.650 And let's see for Hawaii and the so-called green dot ship came by, underscore this dog ship. 01:22.630 --> 01:26.310 So there will be 800 tacos and 200 tacos next. 01:26.560 --> 01:31.030 We are going to apply this night based algorithm and that will be a part of this. 01:32.360 --> 01:34.690 And be Haskell Library. 01:37.380 --> 01:40.080 So from his Gaylan. 01:43.430 --> 01:50.910 Right now, you base, you are going to import this glassine andI. 01:52.970 --> 01:57.250 Let's create the object of this CNN. 01:58.730 --> 02:06.880 Let me say I need to classify it, and next is we are going to do the training on our training dataset. 02:07.610 --> 02:10.770 So it will be classified not. 02:12.860 --> 02:13.970 That will be a fit matter. 02:14.360 --> 02:15.320 Not a trained matter. 02:16.100 --> 02:17.720 So send us your train. 02:18.860 --> 02:20.640 And I understood three. 02:22.280 --> 02:26.260 And let me have any hopes that we'll be a fit. 02:29.950 --> 02:30.440 All right. 02:30.500 --> 02:31.610 So Michael got God. 02:32.280 --> 02:35.330 Let's have a prediction on our training set. 02:35.480 --> 02:43.460 So I'm just going to assign it to why not put any prediction will be on a. 02:43.900 --> 02:48.310 And the school hopes that will be on the center's good test data. 02:49.070 --> 02:54.680 So, Ryan, this data will be our ground truth, whereas Minder's Cooperative will be of a prediction 02:54.680 --> 02:59.450 result, which will be predicted by this Goslee and Nairobi's classified. 03:01.240 --> 03:03.980 All right, so let's find accuracy for this. 03:05.440 --> 03:09.240 Classification model after applying on a testing because. 03:09.520 --> 03:13.430 So that will be from Haski Lone Dog. 03:14.840 --> 03:20.700 Metrics import, let's say accuracy score. 03:21.750 --> 03:23.270 Hey, I'm just going to apply. 03:25.410 --> 03:26.310 Accuracy is good. 03:27.170 --> 03:28.930 So that will be a plus lightest. 03:29.450 --> 03:33.040 That will be a low ground troop level and wideness could. 03:34.610 --> 03:34.810 Right. 03:35.800 --> 03:36.700 Let's just run it. 03:37.360 --> 03:41.260 And it indicates that almost 73 percent accuracy. 03:42.190 --> 03:52.300 That means out of 200 records, if we just multiplied 200 Pequots by zero point seventy three, which 03:52.300 --> 03:54.940 will be close to around one hundred and forty six. 03:55.180 --> 03:58.780 So in one hundred and forty six, that of a prediction was right. 03:59.070 --> 04:03.580 Whereas another 50 for records of a prediction was wrong. 04:04.300 --> 04:04.680 All right. 04:04.720 --> 04:09.480 So that is the whole story of the restaurant review classification dataset. 04:10.180 --> 04:18.370 And in this particular section, we learn about all those pre processing step leiker stamping stopwork 04:18.370 --> 04:24.730 removal, lowering your every single token removal of all those digits, punctuation mark. 04:25.390 --> 04:31.420 And these are the minimum thing you need to do while applying your data to the machine learning by any 04:31.420 --> 04:32.860 feature engineering algorithm. 04:33.250 --> 04:38.920 Although this particular naglfar model also has an a capability to do all those things. 04:38.950 --> 04:40.480 Whatever we did earlier. 04:41.190 --> 04:41.560 All right. 04:41.590 --> 04:45.490 So that is all about the Restaurant Review Classification Project.