Hello everyone, welcome to the course of machine learning with Python. In this video, we shall continue our discussion on machine learning. So what is machine learning the field of study that gives computers the ability to learn without being explicitly programmed. So this was saved by Arthur Samuel. Tom Mitchell said that a computer program is said to learn from experience II with respect to some class of tasks P and performance measures p, if it's performance at task T, as measured by P improves with the experience. So let's try to understand the definition given by Tom Mitchell with a couple of examples.

Example one classify emails as spam or not spam. So here task D is classify emails as spam or not spam. experience would be watching the user to mark or labels emails as spam or not spam and performance would be the number of fraction of emails correctly classified as spam or not spam. So more experienced our algorithm is it would be classifying the emails more satisfactorily into spam or not spam for example to recognizing handwritten digits. So here task is recognizing the handwritten digits. experience would be watching the user to Marco level the handwritten digits to take classes, say zero to nine and identify the underlying pattern performance would be the number of fraction of handwritten digits correctly classified.

So more it is correctly classified more performance. So when do we use machine learning? machine learning is used in the case where there is an intuition that certain rules exist, but we do not know how to express these mathematically. So the machine or computer learns the rule from data. The performance of the learning agent in this case the machine on the computer should improve the based on experience in either of the following ways, the range of the performance is expanded that is machine can do more, the accuracy on task is improved that is machine can do things better and the speed is improved that means machine can do things faster for different kinds of learning. Machine learning is a data oriented discipline.

Depending upon how the machine learning models learn the rules inherent in the data set, there are various types of learning with us. supervised learning in this type of learning the ground truth of the training data set is no that his data set is already labeled unsupervised learning. In this type of learning the ground truth of the training data set is not known. That is the data set is not labeled semi supervised learning. In this type of learning few of the training data sets are labeled and most of them are available. Reinforcement learning in this type of learning The agent will be penalized or rewarded based on the incorrect or correct prediction.

The task of the agent is to maximize the reward, supervised learning, learning under supervision. So let's say teacher teaches a child that this is Apple. And when the child goes back and sees an apple, she will be recognizing that it is an apple. So this is what we call the supervised learning. So there is some kind of supervision or the teacher involved in learning. So the machine learning model is also able to learn from past data and make prediction or classification.

So this past data should be labeled. This is called supervised learning. Examples of supervised machine learning. So let's say we want to predict the blood pressure based on the eat and wait. So here the task is predicting the blood pressure of an ad. Based on his or her age and weight, her here age and weight are two predictors or input variables.

And that pressure is the target or output variable. There are total 242 examples present in the data set. Each training sample is of the form each week and blood pressure, so age and weight here at the predictor, and the blood pressure is the variable which we want to predict. In the supervised learning framework, we shall use a large fraction of our data to teach the model so that the model can learn and predict the rest of the data with high confidence. Let's see another example. Here the task is whether to sanction load or not to a person based on his or her age, monthly income and existing loan amount.

Here two predictor variables are age, monthly income and existing loan amount. There are total thousand liberal samples given the target period is known sanctions alone session could be either yes or no. So, each data sample over here is of the form each monthly income existing loan isn't the feature or the predictors and loan essential is the predicted variable. So, likely the years here also we take a fraction of the total data set as training samples to train our supervised learning model. So, what is the main difference between the examples of supervised learning we just saw So, in the first example of blood pressure prediction the predicted or output variable as us continuous values within a given range, these types of supervised learning is known as prediction or regression problem. While in the second example of whether to sanction load or not the predicted or the output variable is categorical in nature as it as you only find a discrete values in this case, only yes or no.

So these types of supervised learning is known as classification problem. Then comes unsupervised machine learning learning without teacher. It is also called self organization. unsupervised learning is the machine learning task of inferring a function to describe inherent hidden structure from unlabeled data. So, let's say these are few fruits and we want to make two groups out of it. So, one can make the grouping based on the color.

So Group B or group rate, or somebody can classify these according to the fruit type group apple and group tomato. Note that in group tomato, there are both green and red tomatoes. And in group Apple also there are both V and red apples. So depending upon what features we are choosing for grouping the data, this grouping could be different. So natural grouping of objects into two or more groups. That each object fall into exactly one group is called clustering.

The word natural is not quite definitive, it can have different meanings, his different types of clusters possible out of sin dataset. Apart from clustering, there are other unsupervised learning techniques, one of searches dimensionality reduction. In dimensionality reduction, we try to find out a lower order representation of the data without losing much information. So, machine learning learning from data, it can be classified into supervised learning that is learning which teacher and unsupervised learning learning without teacher supervised learning can be further divided into regression and classification. So, regression means continuous valued output and classification, discrete well, output. unsupervised learning can also be classified into clustering that is natural grouping of the data or dimensionality reduction that is finding a lower order rehabilitation of the data.

Now machine learning process. Machine learning is a data oriented discipline. So we need data to start a machine learning process. Usually more data leads to better algorithm, but this is not always true. first identify the task and acquire the required data for the task. Then people assess and clean the data, remove the noisy data, or outliers, if any.

It is a supervised learning task. Then given the pre processed data first bake it into train and test data set, like 70% for training and 30% for test. So this is not a norm. We can use different other splitting like 75% 25% or 80% 20%. Also for training and tests it train your model on the training data set and test it on the test data set. So test data set is usually the unseen data set.

Keep changing your model or model parameters until you get dessert accuracy on both training and test data sets. So these given data, we are dividing it into training and test. Then we'll be selecting a model. We shall train the model on training data and validate the model on test data. If we obtain desired accuracy, then done otherwise, we'll be selecting a different model or we'll be fine tuning the hyper parameters of the selected model will again creating the model validate the model and check for the satisfactory result. So this is in nutshell, the machine learning process.

So thank you. In the next video, we shall begin our journey with Python. So see you in the next lecture. Thank you