WEBVTT 00:00.080 --> 00:01.000 Welcome back. 00:01.040 --> 00:05.640 In this video, we're gonna learn about the types of ML algorithms. 00:05.640 --> 00:09.200 And we're going to deep dive into the supervised learning. 00:09.360 --> 00:17.440 Supervised learning is a type of machine learning where a model learns from labeled data. 00:17.640 --> 00:24.160 Meaning every input has a corresponding correct output. 00:24.160 --> 00:32.720 The model makes predictions and compares them with the true outputs, adjusting itself to reduce errors 00:32.720 --> 00:35.080 and improve accuracy over time. 00:35.200 --> 00:41.360 The goal is to make accurate predictions on new, unseen data. 00:41.600 --> 00:49.040 For example, a model trained on images of handwritten digits can recognize new digits. 00:49.160 --> 00:52.080 It has never been seen. 00:52.120 --> 01:00.960 Under supervised learning for classification, regression and forecasting, classification teaches a 01:00.960 --> 01:04.520 machine to sort things into categories. 01:04.800 --> 01:12.080 It learns by looking at examples with labels like emails marked spam or not spam. 01:12.400 --> 01:20.960 After learning, it can decide which category new items belong to, like identifying if a new email 01:20.960 --> 01:22.640 is spam or not. 01:22.920 --> 01:30.080 For example, a classification model might be trained on data set of images labeled as either dogs or 01:30.120 --> 01:39.080 cats, and it can be used to predict the class of a new and unseen images as dogs or cats based on their 01:39.080 --> 01:45.320 features such as color, texture, shape, tail, and others. 01:45.440 --> 01:51.320 Okay, so the classification means sorting things into categories. 01:51.360 --> 01:52.360 Regression. 01:52.520 --> 02:01.720 In regression tasks, the learning machine must estimate and understand the relationships between variables 02:01.720 --> 02:10.650 in a system by analyzing only one dependent variable, as well as a few other variables that are constantly 02:10.650 --> 02:11.370 changing. 02:11.690 --> 02:18.170 Regression analysis is particularly useful for forecasting and prediction. 02:18.170 --> 02:26.410 So regression, where the goal is to predict a continuous numerical value based on one or more independent 02:26.410 --> 02:27.210 features. 02:27.370 --> 02:33.330 It finds relationships between variables so the prediction can be made. 02:33.450 --> 02:36.170 We have two types of variables present. 02:36.170 --> 02:37.170 Regression. 02:37.210 --> 02:42.050 Dependent variable which is the target variable we are trying to predict. 02:42.050 --> 02:50.450 For example house price or independent variable or features that the input variables are influenced 02:50.490 --> 02:54.010 the prediction for example locality number of rooms. 02:54.290 --> 03:02.850 So regression analysis problem works with if output variable is a real or continuous value, such as 03:02.850 --> 03:04.850 salary or weight. 03:04.890 --> 03:06.050 Forecasting. 03:06.250 --> 03:14.570 Forecasting involves analyzing, analyzing past and present data to make predictions about the future. 03:14.570 --> 03:21.170 As a quick recap about the supervised learning and the steps, the working of supervised machine learning 03:21.170 --> 03:23.730 follows these key steps. 03:24.010 --> 03:26.170 Training data set. 03:26.290 --> 03:27.810 Collecting labeled data. 03:27.850 --> 03:33.410 Gather a data set where each input has a known correct output. 03:33.570 --> 03:38.010 For example, images of handwritten digits like three. 03:38.290 --> 03:39.410 Like this. 03:39.610 --> 03:41.410 Like for example this. 03:41.610 --> 03:48.450 Those are the images of handwritten digits with their actual numbers as labels. 03:48.450 --> 03:57.090 So the label would be number three, the label here would be number three and so on. 03:57.330 --> 03:57.810 Okay. 03:57.970 --> 04:02.290 So every image has its correct label. 04:02.410 --> 04:11.130 Then splitting the data set divide the data set into training data about 80% and testing data about 04:11.170 --> 04:11.850 20%. 04:11.850 --> 04:14.850 And this is what we're going to do in the next videos. 04:15.090 --> 04:17.530 The model will learn from training data. 04:17.530 --> 04:28.100 And we evaluated on the testing data, then training the model feed the training data inputs and outputs 04:28.100 --> 04:36.980 and their labels to a suitable supervised learning algorithm like decision trees like SVM or linear 04:37.020 --> 04:37.940 regression. 04:37.940 --> 04:44.740 So in the next videos, we're going to learn about the algorithms of the supervised learning. 04:44.780 --> 04:50.660 The model tries to find patterns and that map inputs to correct outputs. 04:50.660 --> 04:55.420 Then the fourth step is to validate and test the model. 04:55.460 --> 04:58.180 Evaluate the model using testing data. 04:58.180 --> 05:03.100 It has never been seen before, and the model predicts output. 05:03.220 --> 05:10.780 And those predictions are compared with the actual labels to calculate accuracy or error. 05:10.900 --> 05:13.740 The last step is the prediction. 05:13.740 --> 05:23.020 Once the model performs well, it can be used to predict outputs from completely new unseen data. 05:23.020 --> 05:25.380 Let me show you an example. 05:25.380 --> 05:29.060 Here we have the input row data. 05:29.060 --> 05:31.700 And here we have the labels. 05:31.740 --> 05:38.700 Every element every image here has its own and corresponding label. 05:38.700 --> 05:42.700 So for example this is the image of elephant elephant elephant. 05:42.900 --> 05:45.180 And those are the elephants. 05:45.180 --> 05:50.740 And the label class contains the elephant label. 05:50.780 --> 05:54.580 Also this is true for the cow and camel. 05:54.580 --> 06:05.700 So every element here, every input data has a label gathering a data set where each input has a known 06:05.700 --> 06:07.100 correct output. 06:07.260 --> 06:12.860 Images of animals with their actual names as label. 06:12.860 --> 06:15.300 Then splitting the data. 06:15.580 --> 06:25.900 Divide the data into training data about 80% for uh, 80% for training and testing data, about 20% 06:25.900 --> 06:28.430 for training the model. 06:28.470 --> 06:32.150 Here we have the algorithm, as I told you. 06:32.190 --> 06:40.910 We have as a VM, we have decision trees, we have linear regression and so on. 06:40.990 --> 06:50.990 Based on this algorithm and training data set and the desired output, we are going to use and validate 06:51.150 --> 06:52.550 the model. 06:52.630 --> 06:59.270 So evaluating the model using testing data the model then predicts the output. 06:59.270 --> 07:06.990 And those predictions are compared with the actual labels to calculate accuracy or error. 07:07.150 --> 07:12.110 This is the level of processing the data. 07:12.150 --> 07:15.870 Then we get the output the machine. 07:15.910 --> 07:20.230 Now recognize that this image is for elephant. 07:20.270 --> 07:24.510 This image is for camel and this image is for cow. 07:24.750 --> 07:32.190 Once the model performs well it can be used to predict outputs for completely new unseen data.