NumPy Arrays and Data Analysis

We are going to discuss the NumPy. So NumPy is a linear algebra library for Python. The reason why it is so important for data science is that it covers almost all of the libraries in the PyData ecosystem, which relies on NumPy as one of their main building blocks.

So one more thing, one more advantage we have, NumPy is also incredibly fast as it has bindings to the C libraries. 

Okay, so first we are going to talk about the installation process. So to do that, if you are using a Jupyter Notebook, you just need to do conda install NumPy. By using conda install NumPy, your NumPy will be installed. 

And if you are not using Jupyter Notebooks, so simply go to your PowerShell and just install NumPy. As your NumPy is installed, now you need to do to import it in your program. So to do that, you just need to do import NumPy as NP.

So NumPy has built-in functions and capabilities, and we won't cover them all. Still, instead, we will focus on some of the most important aspects of NumPy like vectors, using arrays, matrices and number generations.

So NumPy array zation surely come in two flavors like what is our matrices and vectors are strictly 1-D arrays, and matrices are 2-D arrays. 

So how to create the NumPy arrays? This is the variable list name off my list, and here I have three elements, okay, one, two and three. And if I just do 'my list' right over here and do Shift +Enter, it will take the output one, two, three. 

Okay, so now the same thing I want to do that I want to create an array and I want to convert my list into my array. To do that, you just need to do NP dot array. Right, so this is the feature which we are going to do, NP dot array. Why NP? Because we have imported NumPy as NP.

So NP is the short form for NumPy. So that's why we are using NP dot array. And inside the parenthesis, we will pass the parameter of our list name, o which is our minus.

Now you can see the changes. First, when we execute the list, it shows 1,2, 3 with the square brackets, but now, it shows within array written over here. The array is also there and the parentheses and inside it, we have the list.

So if you do that with your matrix by taking it one, two and three, four, five and six, seven, eight and nine, so these are the three lists which I have included, nested list. This is a listed list, and if you run that, you will see that, one, two, three, four, five, six, seven, eight, nine.

So this is the execution of our nested list. Right, if I do the same thing same function like NP dot array and write inside the parenthesis my nested list name. So my list and you will see that it has been converted into an array just like this one, one, two three, then four, five, six, seven eight nine.

So basically, it has three rows, right. And it has only a single column, so this is my entire column. So it has been converted into my matrix form, right. 

Let's talk about the built-in methods. There are lots of integrated ways to generate arrays. The first one which we are going to discuss is a range. What does it do? It returns evenly-spaced values with a given interval. 

So I have done that NP dot a range from zero to ten. So I want to include elements starting from zero and will end at ten. So just keep this thing in your mind that whenever you are giving a stop parameter, it will terminate as soon as it finds the ten. It has started from zero, and it has ended in nine.

So similarly, now we have another thing, just NP dot a range. And now I am adding the third parameter, which is our step parameter. This parameter helps you to jump to the step you want to do. In my previous example, you can see that it has been growing up with the increment of one, but now you have given a step parameter, and it has been located with the value of two. Now, it will be growing with a two-step, right.

So we execute this code, then you will find that zero, two, four, six, eight, ten. And see you have terminated at 11. So it has been terminated with one previous value. So your array has been executed, starting from zero and ending to ten. 

Okay, now want to create zeros and ones, you want to generate arrays of zeros or ones. So to do this, you need to do simply just NP dot zero and number of zeros you want in your array, just give the value. So now, you can find that your array has become with the three elements of zero.

Okay, if you want to build a 2-D array or two-dimensional array, we have only one thing, one parameter. So it is defining only your rows, you don't have any columns. But in the second example, we can see that NP dot zeros, five comma five, right. So, the first one is for my rows, and the second one is for my columns.

There are 25 elements now in your array, including 25 zero starting from row 1 to 5 and there are five columns as well. So this is how we can generate arrays of zeros and ones. Similarly, you can also create ones, right. 

And there is another of a 2-D array. So three comma 3, 3 rows and three columns. Now you can find the nine elements, right.  Why there is a function of 0 and 1? Because there are some functions where we need to know whether it is true or false. So 0 represents our false value, and 1 represents a true one. 

Now we want to see about the linspace. So it returns evenly spaced numbers over a specified interval, right. NP dot linspace, what does it mean? Linear space, linspace simply meaning linear space. 

I have given a start parameter of 0, and ending parameter is ten, and the 3 is not my step parameter at this time. Okay, this is the number of elements you want in your array, right. So from 0 to 10, I want only three elements, and it should be equally or evenly spaced. If you execute this, you will find that 0 and 5 are there along with 10. Okay, so the difference between 0 and 5 is equal, and 5 and 10 are also identical.

So your array has been divided into three equal parts, right. For a stop parameter, I have told you that in a range, it doesn't get to stop parameter but terminates one value before. But in linspace, it takes the last value also which we have included. 

So in the array, you can find the same interval and is evenly spaced, right. So this is what our linspace do.

Let's check out with the eye. So the eye creates an index matrix. So if you do NP dot EYE, you will create an index matrix. All your diagonals values are one, and rest values will be zero.

How to create random value? How do you generate random numbers? Okay, there is a module named 'random' in our NumPy. So this random has many features like rand, randn and we are going to discuss it one by one.

To create an array of the given shape and populate it with the random samples from a uniform distribution. So now I have NP dot random. You need to call this random value because the randn is a function of this random module. You need to call, this you can not do just by NP dot rand, you need to NP dot random dot rand 2. 

So what does it will do? It will generate two random values, but keep this thing in your mind that it will generate between 0 and 1. I have created 25 elements by giving five rows, and five columns and all the values are between 0 & 1 interval.

Let's talk about the randn function. It returns a sample from the standard normal distribution unlike rand, which is uniform, right. If you do NP dot random dot randn, it will generate values, which can be negative as well as positive. It can take any values, and it is not that it will only take values between 0 & 1, whether it will be negative or positive.

Let's now discuss array attributes and methods. Okay, if you want to change the shape of your matrix you want to convert your rows into columns or your columns into rows, so to do that, you just need to reshape. 

There are other methods like max, min, argmax, argmin. These are all the useful methods for finding the maximum or minimum values. And argmax and argmin just find the index location of your maximum value or minimum value. 

So in the first example, I am taking ranarr. So this is my array elements, 10, 12, 41, 17, 49, 2, 46, 3, 19, 39. And now I want to find the maximum values out of this array. So ranarr dot matrix name and dot max can do your job. So it is 49 right here. 

And if you want to find the index of your maximum value, just use dot argmax and the parentheses, it will be four. So check the index, zero indexes to the fourth index, and 49 is the highest value over here. So it returns the index of your highest value.

Similarly, ranarr dot min will give you a minimum value. And argmin will find the index of your minimum value. 

Now we are going to talk about shape. So shape is an attribute that arrays have, right. So if you want to check the array, like over here, I have taken an example of this vector one, arr dot shape. It gives a shape of 25 because we have 25 elements on our rows, we have no columns. If you want to change it, you can just use arr dot reshape to 1 to 25. Now it will be converted into one row and 25 columns.

Another thing to know is the type of data which we are having in our arrays. So to do that, you just need to do arr dot dtype, and you will find what kind of integer or data is there. So we have the integer values over here, so 1, 2, 3, 4, 5, 6, all our integer. The dtype is returning 64-bit number value. 

So that is all about NumPy arrays. Because there are some functions where we need to know whether it is true or false, so 0 represents our false value, and 1 represents a true one. 

Now we want to see about the linspace. So it returns evenly spaced numbers over a specified interval, right. NP dot linspace, what does it mean? Linear space, linspace simply meaning linear space. 

I have given a start parameter of 0, and ending parameter is ten, and the 3 is not my step parameter at this time. Okay, this is the number of elements you want in your array, right. So from 0 to 10, I want only three elements, and it should be equally or evenly spaced. If you execute this, you will find that 0 and 5 are there along with 10. Okay, so the difference between 0 and 5 is equal, and 5 and 10 are also equal.

So your array has been divided into three equal parts, right. For a stop parameter, I have told you that in a range, it doesn't get to stop parameter but terminates one value before. But in linspace, it takes the last value also which we have included. 

So in the array, you can find the same interval and is evenly spaced, right. So this is what our linspace do.

Let's check out the eye. So the eye creates an index matrix. So if you do NP dot EYE, you will create an index matrix. All your diagonals values are one, and rest values will be zero.

How to create random value? How do you generate random numbers? Okay, there is a module named 'random' in our NumPy. So this random has many features like rand, randn and we are going to discuss it one by one.

To create an array of the given shape and populate it with the random samples from a uniform distribution. So now I have NP dot random. You need to call this random value because the randn is a function of this random module. You need to call, this you can not do just by NP dot rand, you need to NP dot random dot rand 2. 

So what does it will do? It will generate two random values, but keep this thing in your mind that it will generate between 0 and 1. I have created 25 elements by giving five rows, and five columns and all the values are between 0 & 1 interval.

Let's talk about the randn function. It returns a sample from the standard normal distribution unlike rand, which is uniform, right. If you do NP dot random dot randn, it will generate values, which can be negative as well as positive. It can take any values, and it is not that it will only take values between 0 & 1, whether it will be negative or positive.

Let's now discuss array attributes and methods. Okay, if you want to change the shape of your matrix you want to convert your rows into columns or your columns into rows, so to do that, you need to do reshape. 

There are other methods like max, min, argmax, argmin. These are all the useful methods for finding the maximum or minimum values. And argmax and argmin find the index location of your maximum value or minimum value, respectively. 

So in the first example, I am taking ranarr. So this is my array elements, 10, 12, 41, 17, 49, 2, 46, 3, 19, 39. And now I want to find the maximum values out of this array. So ranarr dot matrix name and dot max can do your job. So it is 49 right here. 

And if you want to find the index of your maximum value, use dot argmax and the parentheses, it will be four. So check the index, zero indexes to the fourth index, and 49 is the highest value over here. So it returns the index of your highest value.

Similarly, ranarr dot min will give you a minimum value. And argmin will find the index of your minimum value. 

Now we are going to talk about shape. So shape is an attribute that arrays have, right. So if you want to check the array, like over here, I have taken an example of this vector one, arr dot shape. It gives a shape of 25 because we have 25 elements on our rows, we have no columns. If you want to change it, you can use arr dot reshape to 1 to 25. Now it will be converted into one row and 25 columns.

Another thing to know is the type of data which we are having in our arrays. So to do that, you need to do arr dot dtype, and you will find what kind of integer or data is there. So we have the integer values over here, so 1, 2, 3, 4, 5, 6, all our integer. The dtype is returning 64-bit number value. 

So that is all about NumPy arrays.