 भी खॉम्वेंच क्यों की लग्टाई बाहँ यी अईज़्ी। ती आदटिए ये जिय्दर थिश्टीं। तो बभाई तुश्डों को यिन प्साथा्चों के और थकों तद्रनें किलिची prophetic क्यों और थकाई। अमने देक्ते है के जो जो मशीन लेरनिग और देटा साँंस यक इस तरा से रेटिद हैं और जो डेटा साँँस को सबजेक्त इसल्फ हैं, यस के انदर, मशीन लेरनिग कितना बोलल प्ले खर सक्ती और कर रही हैं। first of all we will try to understand that what machine learning is machine learning is basically the software programming it is there for almost 70 years but if the software programming language and programming is happening with it it is more than 50 years of history, it is at least 70 years but how machine learning is different, machine learning is a software program but the difference between traditional and traditional programming is that there is a fixed set of instructions which works and you use it if you are working in an accounting department for any type of work you do, there is a fixed method for it according to that software works whereas machine learning is basically the example of the computer that computer scientists are trying to do that whatever work a person does, we should get that work done by all the computers and what we have achieved so far is that we have developed an expert system we have developed a logic, we make different types of software but they don't think like a person because the thinking of a person is different from the method of learning in which supervised learning, unsupervised learning, reinforcement learning this is basically how a person's learning is and the current era in which we are working or the data science which we have discussed there is a lot of role in machine learning because we have so much more data that we need a lot of time and effort to make a system that we won't be able to do or can't do practically speaking for that we design a system which is intelligent enough and the basic intelligence of the data you will also provide it as a data scientist or as a human being and then it works by self-learning this is basically the difference the software design you have to develop in machine learning and the normal ERP system, CRM system or the banking banking system all the systems we have are used and this is basically the difference in machine learning basically there are five steps and the process is the same if you keep your data pipeline in the chain then you process the output or even the data wrangling process or data quality assurance process all these things basically they work in parallel from one place you have always worked as a data scientist as a data engineer, as a data analyst your primary focus is the data set on which you have to work and I just to brush your memories after that you work in this domain your business problem which you have to address accordingly all the things you have the entire data pipeline your quality assurance your data wrangling all these things they work in parallel to each other in this you see the first step is get data these are ETL, ELT, data ingestion basically they represent that after that the next process is this is about data cleaning, manipulation data transformation all these things data wrangling, complete test security all these things the physical or logical data the information managing it, maintaining its quality all these processes you do here what is the difference in this the difference is that if you have a statistical model in inferential statistics in inferential statistics we have seen a different statistical model so we do here that we select a statistical model which can be any of the distributions for the machine learning one of the very basic and the mostly used is regression so first of all your model is a simple linear regression in which there are simply two variables so that is your model and you train it then in this you have when you work on it you use some data for training some for testing if you have 2,000, 5,000, 10,000 daily data available 10,000 records data available then you select one data set initially for testing then you increase it then you train it you will make sure that your data set is a good model that you have selected that is a best fit or nearest to or close to a better fit for your analysis then there are some errors or variances but still they are very good representative of your actual data so based on that when you have worked on test data then you improve it the test data you have then you train it see the results and then you return it then you increase the size of your model the size of your sample this is also very important as you increase the size of your sample the result they are more accurate or they are more true representative of your actual data so when you test and train your data when you see the error the error or some other statistical parameters when you see them then you realize that your desired result is very close so accordingly this model training is definitely a step of model selection when you have selected the model then you realize that your business problem or your data and what model you want so for example I have told you a simple linear regression that you can use this model but we also have complex models which we will discuss in the next module and which will use machine learning types different algorithms different models continue learning about your machine learning process as I have given you a regression example automatically if the data works the software or machine learning normally according to the model we select the data like the source we have taken the data and in the destination it is your database or any folder so it can be in multiple formats and you can use it for multiple objectives in this we do 2-3 major work with machine learning regression, clustering and classification these three things we know that in this regression the simple linear regression there are two variables X and Y one is independent and the other is dependent variable X and Y in classification it is a gender gender is a classification if you want to segregate male and female then based on certain data characteristics you can do it then clustering you have different data sets you have a large population it is available you want to see income group it can be a cluster education level it can be a group or residential area it can be a group basically clusters you will remember that when we were talking about statistical model fitting in that we saw when you plan your model where are the populations where are the clusters we are just seeing how many populations are in which cluster we are not going to further detail that how many male and female gender education income group we were simply seeing in which area they have facilities available public transport, private transport or whatever other means are available so basically we achieve this with clustering and in clustering again the simple selection that is very very critical because the statistical analysis you do the analysis you do that is based on statistical data and the statistical model you apply, the theorem or distributions everything is directly whatever your performance is that is directly dependent on the quality of the data and that is why it is important that when you select samples that is a subject that you have to do simple random sample select you have already clustered and it is called a stratum so you have to select samples so it all depends basically on how you select samples after that you have to apply normal data engineering and you predictive or whatever you do whatever your model depends on that it is very very important