Python for Data Science #5: Mastering Data Visualization with Matplotlib, Seaborn & Plotly

Key Takeaways

So you all know that we are building our portfolio in the world of Python especially for people who are new to the tech world and people who are aspiring to be data scientist, data analyst, gener. So here we are building a portfolio in the world of Python and today we are going to discuss about the content where we can use Python to deliver visualizations, stories, you know your insights. This is where the fun begins where your data gets converted into a beautiful art. So to do that, Python allows various packages. As you know in our previous sessions or in recordings you might have seen that you know we have used Python packages like pandas, numis, stats, math and similarly we have something called math plot like cb plotly. What these packages do they allow us to play with our data convert those boring data into beautiful charts that conveys an amazing story. So today you and I going to learn about these charts and visualization in Pyth. We are learning about packages that can help us convert our data into beautiful visuals. So for that I'm using package called Mplot li which again is a wonderful package that can convert our data into charts. Right. Numpy that can deal with mathematical operations. So I'm calling an alias on mattplot lip as plt and an alias on numpy as np. So you must be wondering what these data points are going to do. Right? So if I show you I add one more cell here. Add this and run it. Right? So you see numpy. Space has created numbers. How many of them? 100 of them between 0 to 10 randomly generated numbers. Then I'm passing those values to a function mathematical function sign function and saving them in x and y variable. So on x-axis I have 100 numbers starting from 0 to 10 and on y-axis I have sign of that like sine of 0 sine of 0.1 s of 0 0.2 like all those values. Then using mattplot lip I'm defining a figure right? You see this figure it's of length six and four height. Then I'm saying plot a chart with x data on x-axis and y data on yaxis. Label it sin x. You see this label sin x and the color will be blue. Title of the chart is baseline plot. X label is X-axis, Y label is Yaxis, legends are drawn here and then convert it into a grid. As you can see the tabular form and print show. So when I run this cell, it generates a beautiful chart for me. Now imagine you have data like age versus salary, engine size versus speed of the car. I can pass those values here and convert that into a beautiful visual. That's not the end of it that we have more. Then let's make complex charts. What if you have some you know intensities? You have multiple dimensions of data. So we have an advanced package built on top of mattplot lip called seabon. Seabon is again a package that is being imported as SNS. Seabon is a way more complex package. It has its own data sets also. As you can see, I have imported a data set. You remember in our last class, we converted our data into uh data frames. Right? So, this is my data frame. Pandas data frame. You remember head tail, we talked about this in the last recording. So, what I'm doing is loading this predefined data from my seaborn package load data set called tips. Then I'm plotting three charts. A distribution chart, a box chart, and a heat map. For each chart, I'm writing four lines of code. Defining the area that I want for the chart. It's a histogram with data coming from tips data frame and a column name total bill. Kernel density estimate true and the color is purple. Title of the chart is distribution of total bill. Let's look at this chart. It's a histogram. We all agree. Kernel density estimate is this line which gives it a normal distribution curve. You know how the data flows and the color is purple with the title as distribution of total bill and plt show means it is telling my compiler to print that chart. Second again a new figure of length and breadth. It's a box plot. What's a box plot? It helps us identify the minimum value outlars to a maximum value with outliers and it gives me a range of my data. Helps me identify the median my uh you know uh distribution of my data whether it's right skewed left skewed we can identify a lot of things with it. Generally it is used for outlier detection in machine learning. So on x-axis is the number of days and y-axis total bill data frame tips and pallet is the color density and uh the format and all of those things. Title remains box plot of total bill by day. Look at this chart. On Thursday the bill starts roughly from $8 till goes till $40 plus. On Saturday the sale comes as low as $2 and it goes above $50 also. So see how visually I can compare my sales. Obviously the sales are good during weekends but the range is high. There are certain transactions which are very low and certain transactions which are extremely high. That's the beauty of your visualization. And the last chart is correlation. Correlation tells me how one column is correlated with other. Correlation and statistical things you will learn in the stats classes. For now consider it when something grows together or in a reverse fashion correlation can capture that thing. As simple as that. If I keep on increasing the engine size the mileage will come down negative correlation. If I increase the engine size speed increases of the car positive correlation. And then there is a case of no correlation. But those things you will learn in stats class. So I'm using data frame calling the correlation function only numerical columns because correlation is for numerical data and pass this from a heat map you see this thing annotation true means give me the degree between correlation values is between minus1 to one okay and these are like color and the format the format of the text so you see the heat map how the numerical columns are related to each other are they positively related negatively related or no correlation. In our case, everything is positively related. So that's the mattplot lip and seabbot. Then we can do some formatting. You know that's not the end of it. We can use some predefined styling just like in web development. Once a basic website is done, you improve it like you know you add some fancy visuals and all that thing. So here also I'm adding those things. I can I can plot a line chart one on top of other. You see two line charts sine and co cosine. You see the sign curve and cosine. So I can plot two charts on one chart also. Sometimes you have to compare sales in Germany versus sales in France. Right? So I can do that also in the same chart. Right? I can even add annotation on top of it. Look at here the second chart. You see this? I'm adding an arrow button. That also I can do using the annotation keyword. It allows us sometimes in in in presentations you might have seen you know people are pointing out okay this is the month where the sale went down. This is the month where the sale goes up. This is how through Python you can add those fancy visuals on it. That's not enough. You know the real world has amazing data in terms of time series like stock market, gold prices, GDP, petroleum cost is uh crude oil prices and whatn not. So I'm creating a date using pandas which will start from 1st January 2024 till it will be f March 1st March and the frequency will be day like every day second 1st January 2nd January 3rd January till 1st March then I'll randomly assign a number random number for each day which will be a cumulative sum like the previous number add on random number to keep on adding it and convert that these two into a dictionary. Key value key value. This is a key. This is a value. This is a key. This is a value. Pass that dictionary into a data frame and convert this into a data frame. Pandas data frame. Now I have created a random data. Converted into a data frame and then plotting a chart. See time series data with random values. Cumulative sum means whatever the value was there now I'll add it. It could be negative also I'm generating a random number here and eventually a time series data has been generated. I mean I can do this through macplot also and I can do this using seaborn also. See, so as of now you have the capability of converting your boring data into amazing visuals using packages like mapplot lip or seab ball. This is what we use Python for, right? To make our life easy, right? Now you must be wondering all these charts are dead, right? I mean they they're two-dimensional non-interactive charts. I mean they are beautified. They are telling a story that serves the purpose. But can I get an interaction? What if I put a hover over over my mouse here and the values come alive? Don't worry, we have a solution for that too. There is a package called plotly which makes interactive charts. So I'm importing a package called plotly px a plotly chart pio. I'm rendering this notebook to make sure it is able to interact and then building a scatter plot data frame name x-axis yaxis color size and title everything in one function as a parameter to my scatter plot and I'm doing show look at this chart see if I hover over it tells me it's Sunday bill was $20.69 $69 tip was five size was five. Isn't it amazing? Right? So you can make a interactive chart like this. Similarly on time series look at here the previous time series was dead right? I couldn't figure out I have to eyeball here like what is this value? I don't know. But look at this chart. Simp uh m plot sorry plotly line chart data frame name x-axis yaxis title of the chart and it is making a interactive chart. Isn't it fun? Just imagine you convert your data into such visuals and go to a presentation use them in slides and tell them what is the business doing good if their business is doing bad and how amazing it would be through coding. And that was the agenda of our today's session. Right? So I hope you enjoyed this thing the world of MACplot lib seabbond and plotly right. So go out there play with these notebooks. You will get the access to these data. You can run them on Google collab jupyter notebooks or visual studio with python interactive enabled. And if you have any questions, queries come reach out to us. We'll be happy to, you know, teach you, interact with you, grow with you, and collaborate with you. Until then, I'm Jent Mahara, and thank you very much.