I mean, if your model takes ages and ages days, weeks maybe even months to process some data and gave you a result probably not the best model and you’ll probably want to either adjust the data set in so it or you want to adjust some of the premises as well as finding stuff like how to store our model as in the beginning values that we put in can be a challenge as well as typically will provide some initial value to these premises and then allow our model to adjust these values to find the optimal solution from best digital media companies in Mumbai.

However, it’s kind of difficult to understand how much of a challenge it can be without seeing the actual model right in front of us. As a quick example. So we’ll save a further discussion for this when it comes time to actually start building up our simple linear regression model. So that’s the basics of building up our models so once we have the model built the second step will be to obviously run our model for digital media companies in Mumbai. For social media agency, what I want to do this in a few different stages and the first stage will be to train our model so once we’ve trained it we’ll run it on some testing data and then once we’ve done that and we’re happy with the results we think our model is accurate enough then we can go to put some real data and go about using our model in the real world.

So basically once the growth is built we run what’s called a session on our graph and that will output any expected results or how accurate the model is what the expected values are and so on and so forth. Now when we run it obviously will need some kind of input values to start off with and some kind of expected outputs. Now, in digital media companies in Mumbai, this is mostly used in our training data whereas when we’re going to test it we’d simply just put in the inputs and then it would give us some outputs and we assess the validity accuracy of those results as well we’ll want to specify how many parks we want to run as a model on which is essentially how many times we want the whole model to run. And also want to specify stuff like the learning rates which is kind of how values are going to change over time and how quickly they will change again this is a bit of an art in and of itself because if we run the model too many times it’s just going to take too long to calculate the data. But if we don’t run it enough times then we’re probably not going to get accurate results. As digital media companies in Mumbai, it is similar to trying to find the optimal learning rates because if we learn too slowly it’s going to take ages and ages to run again if we learn too quickly then we might get a slightly less accurate model. So although it’s not difficult to use tens of so actually makes things really easy again by providing these high-level API.

The challenge comes in finding the most efficient way to go about using it to solve your problems as well as finding enough a four-hour training and testing datasets can be a bit of a problem particularly if you are using stuff like images. Luckily there are massive libraries out there some of which we gain access to thanks to tends to flow and huge datasets that already allow us access to a bunch of already trained data or if we’re building something completely new novel then we have to train the data ourselves. So although building the model can be the toughest part training the model depending on how much data you using can be the most time-consuming part. But I’m not giving you all these warnings to kind of scare you away from this quite the opposite. I hope to treat you guys and get you guys thinking about how you might go about building your models how you might go about training them in keeping all of these ideas in mind as we go to build our models to hopefully solve some interesting problems. And speaking of the models for digital media companies in Mumbai, I’m actually at this point going to assume that you know at least some of the basics about some of the simple machine learning models that we might use. Now we’re going to cover mostly linear regression we may move into some of the other ones depending on how much time we have. But if you completely unsure know nothing about machine learning and once know a bit more about some of the mechanisms behind these machine learning models I do recommend that you go and search for yourself as these tutorials are more about how to use tends to flow and incorporate it into your programs than about machine learning as a whole.

Now that being said I will explain a bit more about linear regression in a few sections when it comes time to actually build a simple linear regression model. So now just think about it as being basically a way to estimate a line that fits some data. Okay. And then we’re going to run our model to basically optimize this fit. So essentially what this will do is again like I said linear regression is just using our basic question for line y m x plus B and then trying to find the optimal values for the slope and the intercept. And this will help us to determine a line that best fits some data. So that’s really what it does. It basically we see it in the inputs and the expected outputs and it trains this line to kind of move around until it finds that model that minimizes the differences between the X. The expectations and what’s actually being outputted