 today's welcome to module 184 of Introduction to Data Science course and in this we will discuss the other aspects of ethical issues we do understand that the algorithmic bias that is one of the very very important part because we saw that when we collect data and put a statistical model on it there is some variation of standard deviation in it there is a little bit chance of misrepresentation not exactly misrepresentation but in accuracy there is some, whatever your predictions are there can be some shortcomings because machine learning or as a data scientist our focus is that is basically data science and your role as a data scientist and we are talking about the ethical values within these couple of modules this segment so the algorithmic bias that is because of the algorithm look what is the algorithm algorithm is basically your statistical model which you use for data collection, cleaning, completeness and then you compare it to different data sets using different techniques and then the bias created in it we know that because in statistics bias is inevitable it is there, it has to be there otherwise the 100% truth is the same and we do not know about the future, nobody knows in fact algorithmic bias is related to many things it has a lot of applications but where you have unstructured data structure data and unstructured data these are the things we have read in the introduction you will read them in more detail in your other subjects but where we talk about graphics we talk about images so there is a small mistake in the algorithm which it can be very very harmful but there is another aspect of this that is discrimination you see in the world where we live we have seen even in the most advanced countries of the world that the ethnic bias the racial bias that plays a very very I would say bad role that this is such a thing we have seen in very advanced societies where somebody forgets about third world countries or poor countries but even the advanced and the most rich countries in the world that some people discriminate their race, ethnicity or religion this is one of the biggest crime in the human history if we talk about morality this is the biggest crime there is a facial recognition there is a lot of solution in the market where you can recognize a person gender is the easiest thing to recognize age group, smiling face face expressions but at the same time you color your features and then you compare with them this is your algorithm which compares that if this person does this he is of African background or Asian or Caucasian Caucasian basically all the white people in the world they are called Caucasian race so in the same way it is African Asian so unfortunately there is a bias in this so this is a very big use case if the data scientist if he himself is Caucasian or he himself is African then this bias in his mind can come towards others and whatever algorithm he has developed somewhere in it nobody knows but there is an element of bad intentions and it can be ethical which stops you from doing quality work now the algorithmic bias now the algorithm is the combination of data is definitely input but the whole algorithm is a software a solution and maybe a component of a solution its components first of all it is a statistical model which you have selected after that you have your algorithm you have written it in software language you have used python you have used r you have used java whatever you used now the script of it because this is something you are predicting the script you are writing even a small mistake maybe not a mistake because some logical mistakes are like this these are logical errors which are not able to detect your quality gate type software nor are there such things which you are testing you do something throughout your software delivery life cycle so you don't know and basically these things this is important to understand the statistical model we have selected is the best fit for the problem to address because if a statistical model corrects one thing then it cannot do the other we will see that it is a poison distribution t-scare distribution if we talk about the average then the mean median mode these are three things so average in layman terms but everything represents something different or with a different aspect so you as a data scientist have to make sure that if you use the right statistical model or tool whatever you write in your code it has to be effective and efficient so on and so forth this will basically minimize the bias and that will reflect decision making whoever decides on this they will make good decisions and its impact on any human life if we again I will just give an example of health care if it has an impact on human life atleast that will be negative impact will be minimized or your positive impact that will improve thank you