# How to make models based on any scenarios?

A hypothesis we need to create Invested Yes So the hypothesis essentially is that now the samples that I have identified our representative enough off my population As you learned that the results of my test our representative off my population That’s the hypothesis that we go with So there’s a hypothesis that we need to validate it So you calculated your mean standard deviation scores variances everything based on samples Now you need to ensure that these are representative Enough of your populations.

This is what inference going on So example the population will tell you to try and extrapolate So you’re going from an unknown scenario One on one scenario on extrapolating findings influences from your known information during unknown information Is Carl s hypothesis testing That’s what the information is there sticks is all about So the inferential star at a six-minute defensively hypothesis Testing right.

Yes, Inferential statistics Predominantly their own help with assistance stick Should anyone anymore tarts questions clarifications that you have in mind at this point One question Graham seven During the session we see usually in the examples and all the population mean And the population Saturday vision is even which in practical use case I believe that getting a population means aside and if it is not easy right So for instance that this actually works Yes So in another video a bit and then talked about when you don’t know population statistics how do you go about for example in standard deviation you do not know So then use a sample statistic that succeeded I just want says that test.

You’ve got a very good point I think you’re two steps ahead But let me address the question because you asked it a majority of the times you would not know the population today Emitters in that scenario with this Ah Carol that the distribution that you’re trying to obey Can you be sure that it is a normal distribution Why we need a normal distribution We’ll come to that when we talk off Centre LTD Um but can we know that it is a normal distribution We’re not sure Then whatever you calculate for the metrics on that particular one and you identify the intervals essentially further particular one you are not sure what you’re going to calculate so you have to extrapolate this one.

What Basically what you essentially do is you become a little more guns away to with your estimations when you work I would sample so because the populations it’s more get normalized So if you deal with ah let’s say is totally off X for a population when you take a sample X maybe representative explained at the representative So you try to be more stringent More this thing, For example, let me say this the same example that we talked about the fruit juice your Gallic leading for the sweetness.

The sample that you guard for all the samples together you’ve got essentially is 8.5 is a sweetness rating that is out there withstand the deviation of one point fight So what does that mean What is the value that can oscillate between the sweetness of the juice plus-minus 1.5 center deviation from the mean So how much That would be on one and it would be seven on the other end It could be 10 so the value should be between seven and 10 Predominantly that’s what your ah estimates should be at the end of the day but that is for the sample Any number of samples you pick it could be like that But what That 8.5 could be for your population We do not know and we may not be You don’t know Maybe we’re not Be.

You don’t know population statistics Then what we try to do is increase the impact of standardization that you’re trying to do so you can start a higher standard deviation You can start a lawyer means so there are ways to do that for your standard deviation formula that you calculated end of the day on There is another time that comes as degrees of freedom So where you divide your rest standard the error for your mean that you calculate is you remember that you know the formula to stand a letter that comes across in the radius is the standard deviation by squared off in a square root of friend stuff and you two in minus one So you increase the value of that particular one so that you do that.

The denominator you’re trying tend to increase that in minus and there’s a lot of factors called degrees of freedom Another piece is coming to play there on because by doing some statistical approximations you tend to move from the distribution to Ah their distribution normal distribution Because of the moment you make these edits your tedious mission tend to behave like a normal distribution.

Then you feel that the sample statistics that you got a the mean our ah the other metrics like standard deviation Raelians Any other things that you get tend to be a hill for the normal distribution on represent a population correct so lets more So let’s think about a gentleman from one question before one here for example that you gave off like taking samples from the for against like Are we saying that we heard like five populations and five different samples No no you cannot have five population That’s a different scenario that that’s one of the important problems that will solve in the hypothesis testing on when we talk of different type types of hypothesis tests that we do we will end up with the particular sample on another sample.

We have two samples that you’re working around with but the bigger question that you start tackling is whether these two samples belong to the same population All belong to the different population That’s an important question for us to tackle We need to understand that dimension also when we solve this problem So if the hypothesis is so there is something called as a paid test Independence distressed off independent samples That stuff paired samples When you talk off a paid sample it essentially means that it’s the same population but at a different for himself Alexa for example you launch your product What is the feedback about that particular product in the 1st 1 week on.

It’s the product is new so people just tend to get a view of that product and they would not have got from exposure to that So they give the initial perceptions Now the product is being used by the customers next For example three months at the end of three months you go to the same sort of people and then you captured that perceptions So now that they spend more time with their particular product.

Are they feeling happy with the product Are the most memorable with that product are now that they spend more time They understood the problems that issue challenges with that product and they’re not really happy So what is that It’s the same population I do But I get different measurements your understanding what I’m coming from Yeah I get that one So I mean before we let’s say going back to your problem Ah Forest We were planning to learn the product rates so I am basically dividing the whole country into five different parts and each of which has its own I put a statistic right But at the end of the day if I launch nationally I would also get a sense off in South What is my average scores That would be like in art.