Data Analysis Using R Programming 2026 | Data Analytics With R | Data Analytics Course | Simplilearn

Key Takeaways

[Music] Hey there, welcome to our data analytics using our full course by simply learn. In today's digital age, data is everywhere. Every click, purchase, and interaction generates information that holds valuable insights. But raw data on its own doesn't mean much until it's analyzed and transformed into something useful. That's exactly where data analytics steps in. In this course, you will learn how to use R, which is one of the most powerful tools for data analysis to uncover pattern, create impactful visualization, and even apply advanced methods like time series analysis. We'll also break everything down step by step. So whether you're just completely new or looking to sharpen your skills, you will feel right at home. You'll not only explore the core features of R, but also see how it compares to Python, discover the top tools in the field, and even get practical tips to ace your next data analyst interview. And by the end of this course, you will be equipped to handle real world data challenges with confidence and start building a strong foundation for a career in the growing field of data analytics. So, let's dive in. Also, if you are interested to supercharge your career in data analytics, the professional certificate program in data analytics and generative AI by ENIC Academy, IT Gojarti is the perfect choice for you. This 11-month live online course offers interactive master classes by IT Goarti faculty and industry experts combining cuttingedge tools like generative AI, charge, Python, Tableau, SQL. You can experience hands-on learning with real world projects and immerse yourself in the campus environment with a special session at IT Gojhati. Plus, you'll be earning executive aluminina status from IT Goarti and IBM recognized certifications to stand out in the job market. So, what are you waiting for? Hurry up and enroll now and you can find the course link below. Welcome to data analytics with R from simply learn. So, are you excited to begin with today's session? Yes, you are. Okay, great. I'm excited as well to be a part of this session. What are we going to cover in today's session? We will cover couple of topics. We will begin with understanding the basics. We will focus on why data analytics and the importance of data analytics and then what is data analytics. We will also focus on understanding the life cycle of data analytics and I'll also try to give you some realtime examples so that we can relate these concepts with the real time examples or scenarios and then we will understand the types of analytics and what do they mean. We will also focus on the benefits of using R. For this session we will be using R studio tool. Hence we have to understand why we use R for data analysis and finally we will perform some hands-on exercise. In today's session we will only focus on how to import the data, how to structure the data and also how to create some beautiful visualizations by using R studio. All right. Python and R are top programming languages that are in demand in data science community. Both have their own pros and cons. Therefore, it is vital that you understand the crucial differences between the two programming languages and decide which one is the best fit for you. After watching this video, you will understand which language is better for you in terms of different parameters. Before unwrapping today's topic, I request you to subscribe to our channel and press the bell icon to never miss any updates from Simply Learner and also get notified every time you get a similar video. Now let us look at what are we covering today. We are covering introduction to R and Python. Parameters in R and Python. Best tools and libraries offered by Python and R and conclusion to the topic. Introduction to R and Python. Let me ask few queries regarding Python and R. Our language is super superficially related to Python, C, C++, Java. Please leave your answer in the comment section below and stay tuned to get the answer. And another question is which one of the following is a Python file extension? P dot Python py and pyt. Please leave your answer in the comment section below. Coming to introduction to Python and R. R is a statistical programming language and environment that integrates statistical computing and graphics. R is powerful and stable software. Python Python can also be called as a generalpurpose programming language for data analysis and scientific computing. Python can be considered as the best player in machine learning. Python is an expressive language with many built-in function. Both are open-source software and platform independent and they are platform neutral and also compatible with all major operating systems including Unix, Windows and Mac. Next we will be covering different parameters. We will be covering learning preferability, mathematical fundamentals, speed of both languages, visualization and graphics, data handling capacity, demand, community and customer support, employment possibility in both the languages. Let us cover it one by one. First one is learning preferability or ease of learning. Python is renown for its ease of use. Python's notebooks offer excellent tools for sharing and documentation despite the fact that there are currently no GUIs for them. Programmers find R as difficult language as a beginner. This implies that the programmers must devote a significant amount of time to learn and comprehending our coding. Coming to mathematical fundamentals required. Coming to Python, understanding descriptive analysis is very important. In layman's terms, descriptive statistics often refers to the process of explaining using certain representative techniques such as charts, tables, Excel files, etc. Python statistics is a built-in library for descriptive statistics. If your data sets are not too big or if you can't rely on importing other libraries, you can use Python. On the other hand, R requires basic statistics. From basic statist statistics, what I mean is mean, mode, and median are the terms used most frequently in basic statistics. It is referred to as measures of central tendency. Probability statistics plays an important role in handling various types of probability distribution. It includes binomial and normal distribution. Next parameter is speed. Python is an interpreted language with dynamic typing. Python always executes slowly because the code is executed line by line. Compared to MATLAB and Python, RS ourr language is significantly slower. Our packages are substantially slower than those for other languages. Now that we have covered speed, coming to data visualization and data collection in Python. When selecting data analysis tools, bit visualization are crucial and Python has some incredible visualization tool. In Python to large and varied scatter plots using regression lines, we can use ggplot 2 and ggplot tools. Compared to raw values, visualized data is easier to comprehend. Therefore, R has many packages that offers sophisticated graphic features. In R we can use in R we can use tools like M plot lip sabon etc. Data handling capability in both Python and R. The new releases in Python have resolved the issue with the Python packages for data analysis. R is useful for analysis because of the abundance of packages, accessibility of the test and benefit of employing formulas. However, simple data analysis can also be done using it the need to install many packages. Crucial part of parameter that is tools and libraries in Python and R. As a Python developer, one needs to be wellversed in the best libraries because Python has a lot of libraries that have many different uses. Libraries like TensorFlow, Scikit learn, NumPy plays an important role in solving many Python related problems. Libraries perform a wide variety of task in R that are very beneficial for data science operations. Example for that is depier, bioonductor etc. Community and customer support support index offered by Python and R. Compared to R, Python has a larger community. For assistance, we can contact www.python.org. For any queries regarding Python and help you can support uh you I repeat you can visit support.realpython.com. For any help and queries R offers you with R studio community. R provides assistance through its official website. For queries and community related issues we can contact www.heenpro.org. auto. Next is job opportunities in Python and R. A recent survey from indeed.com predicts that at least 55,000 Python jobs in the USA with exponential pay rates are available. Big tech companies like Google, Amazon, Twitter, Facebook requires Python developer to handle massive amount of data. Position provided for a Python developer is software engineer, data analyst, data scientist and many more. Career in R is an excellent job opportunity for you as a beginner. Big tech companies like Google, Twitter, Facebook are using R. Position provided by companies as a R developer is data scientist, data analyst, data visualization analyst etc. Moving on, let us wrap up an important topic which language to be used between R and Python. There is no right or wrong way to study both Python or R. Both are in demand skills that will enable you to complete almost any data analytics work you come across. It ultimately depends on your background, interest and career objectives that which one is better for you. But compared to R, Python is easy to learn. Let's compare its strength and weaknesses. It is used to handle large amount of data. Python performs non-stistical functions and it is best suitable for programming. However, Python is better when it comes to coding. Whereas R is used in data visualization graphics, R is a widespread language in the statistical community. It is used to accomplish many mathematical task. So before concluding the topic let me answer the query that I have asked regarding R and Python. Do you guys remember the question? The query was R language is superficially related to which language. So the answer for the question is C language. Next question was which one of the following is a Python file extension? It was an easy question. Answer for that question is py. So firstly let us understand why data analytics. Can somebody here tell me why data analytics and why is it used in organizations? So let us understand why data analytics and what is the use of data analytics. As we all know that data is growing exponentially year over year. It is collected and it is also available everywhere. Data is no more just available in structured format but it is also available in semistructured and unstructured format. I'm sure you would have come across this term called as sentiment analysis pretty often, right? What does that mean? Can we perform that in data analytics? Yes. Right. We use some natural language processing models. Try to identify what are the good reviews, what are the bad reviews spoken by the customer. Correct? So we try to classify the text based on the good, bad and the neutral. So that is what sentiment analysis about. So for that to perform that activity we have to also do some data analytics. Now that companies have realized the importance of these informations not just structured but also unstructured data format. The companies have started utilizing these data to take some crucial business decisions which can boost their business and also which can increase the efficiency of the business. So now that the raw data is accessible to the organizations, it becomes very important that the data is also stored. Well, I'm sure you all have pretty much heard about data warehouse. In the last few decades, the trend was mostly on the data warehouses, the business intelligence tools. So, the data warehouse used to collect these data, pre-process the data and also filter the data and make it available in a structured format for further analysis. However, now that is not the scenario. A term coined as data lake is available and many firms are utilizing data lake because it is a central repository which stores the raw data in form of structured and unstructured data. And now let us take a scenario. Let us choose one of the participants here Mark. So Mark has recently joined an organization as a data analyst or in essence a data scientist. The business connects with him and says that Mark we have a business problem for you and we expect you to provide us a data analytics solution. So Mark sits with the stakeholders and he listens very carefully to the business question. So the business question says that we have couple of products which are performing really well in the market and we see higher sales and some of the products are just not catching up in the market and we experience lower sales. Can you help us identify what are the factors driving this higher sales and the lower sales? Now Mark has to think with the data scientist mindset and be prepared to ask them right questions. some questions such as do you have the price of all these products available in the database and can I also know what is the duration of data availability and also Mark can ask some question such as do you also have some features of these individual products already captured in your data base. So these are some of the interesting questions that makes sense to the business and the conversation continues which also means that data is not only information. Data analysis is about unlocking insightful informations from this raw data. And hence data analysis plays an important role in discovering insightful information, asking questions or answering the right questions and also predicting the future or the unknowns. And to perform all these activities, we use data analytics. So are you all with me so far? Are we on the same page? Yes. Great. Now the question is what is data analytics? So can somebody tell me what is data analytics? So let us see what is data analytics. We understood why data analytics and the importance of it. But to perform any activities there has to be a process right. So data analytics is a process to extract meaningful insights from data. Now let us continue with the scenario of Mark. So Mark now understands the business problem and he has also started asking some relevant questions to the stakeholders and he has also got the answers in return. He may start thinking about what could be the suitable solution for this. He may have to perform some exploratory data analysis to unlock some hidden patterns to identify some correlation between the variables and to also know which are the key variables in the data set and he may also have to view the market trend which means he may have to see that how the sales has been performing across the years or across the months. There might be some insightful information there. He may see that the sales has been growing exponentially for some of the products and some may be volatile. Some may have some seasonality pattern or a cyclical pattern etc. And also he may have to focus on the customer preferences via the customer reviews and do some sentiment analysis. So now let us understand what are those life cycles of data analytics. We will begin with the discovery phase. This is the first phase. Now that Mark has understood the business problem, he will also start focusing on identifying the resources that is the data resources. Some may be internal data resources that is uh available within the firm. could be some transactional data and some external data sources maybe via web scrapping identifying and capturing some um competitor price on the products. After gathering all of these right data, Mark will focus on data preparation which is the next phase. Now Mark either individually or along with the team will start focusing on the data preparation which includes data wrangling which means cleansing the data imputing the records if there are any missing values or you know he may also go ahead by removing those records if they're not required and also doing some exploratory data analysis which can include some statistical analysis like looking at the data distributions, understanding the summary of this data distribution at individual variable level, doing some biariate analysis and also trying to you know figure out which are the important variables that might be required for the model building phase. After performing all of these EDA activities which also include some visualization, Mark will now sit with his team and try to identify the suitable models. The suitable models could be simple statistical techniques or it can also be some machine learning models. So let's say that Mark and his team has identified some five models that can provide the required result and out of these five models they will filter down and they will prioritize only three models. Now there are only three models that Mark and his team have finalized. After this they start focusing on model building activity. Now for model building activity they have a data set already in place. So this data set will be split into training data set and test data set. It's not only training and test. We can also do some validation in between training and test. But here let's focus on training and test data set. He will separate 75% of the data as training and 25% of the data as test. Now if your question is that why perform this activity of splitting the training data set and test can't we just go with the one single data set? What happens if you just utilize one single data set? Let us say that you have used the original data set and uh also executed this data in one of the selected model and you will also observe the accuracy. Let's say the accuracy return is about 98%. 98% is a very good accuracy percentage and you may also be overconfident because of this. That may be a case due to overfitting. Now what will happen when you add some new records into this data set and you rerun it? The executed model may not return you the same accuracy what you had seen. The accuracy might be 72%age. Now that's not fair, right? To avoid these overfitting issues, we ensure that some new records are tested separately. So hence we locate 25% of the data to the test data set and then we predict this records the unknown records which is located in the test data set. we predict them and then we test the accuracy of training data set and the test data set and we make a comparison. Now let us say the accuracy result of training is 98% and the test is 97%. In this case we can say that the model is performing really well. However, while executing the model there are certain things that has to be considered. for example, inclusion of the parameters, tuning the parameter which also will execute the optimal results. So this is very important. Now let's say after performing this model building the time comes to analyze the results. That is the next phase. Now the team will sit and analyze the result and they will notice that out of the three filtered models only two models are returning excellent accuracies. They will sit with the business team and they will also explain them the result and what are the activities that they have performed to obtain this result. Some of the stakeholders may be technically savvy, some of them may be non-technical people. So it has to be very important that you also communicate these results accordingly. All right. Now that you have the results, you will also gauge them based on the business objective which was developed in phase one. Looking at the results of the two models. Now the business might select one of the model and say that okay this particular model seems to be returning some right information and it also appears valid to us. Let's go ahead with this one model. So finally the result and the model needs to be operationalized and that is where the team will start documenting the business problems the steps that were taken for executing the models and they will include all the codes and the findings and finally they will implement this model so that uh the business can view the results and also utilize them for strategic decision making at your firm. form. So, are we good so far with the understanding of the life cycle? All right, great. Now, let us focus on the types of analytics. What are the types of analytics? Can somebody tell me which are the types of analytics you are aware of? Okay. Predictive analytics. Great. Descriptive analytics. Good. All right, good enough. Now let us focus on this example of Google maps. So as we look at this particular Google map, we understand that the blue color route is nothing but the root direction from Sacramento to Floren. And also we see a display of the duration estimated as well the distance to travel from Sacramento to Floren as well. We see another route here that is gray colored. This is a substitute route or the connecting route just to avoid the traffic which is in orange within the blue color route. So the gray colored road as well shows us the estimated duration to travel as well the distance. Now let us understand what is descriptive analytics and why do we focus on this particular map example. As we understand the root map, the estimated duration as well the distance to travel via the blue color route as well the gray color route. This is one way of understanding descriptive analytics as in what is happening. But there is another way of understanding descriptive analytics that is by focusing on summarized past data. And this is a descriptive analytics. We see that what had happened in the previous year. Now let us focus on predictive analytics. What is predictive analytics? This type of analytics looks into the historical and present data to make predictions of the future. What does this mean? So Google has already suggested the best route which is the blue color route to travel from Sacramento to Floren and the duration is 18 minutes and distance is 9.7 mi. Let's assume that Google map has already collected historical data of this particular route and based on the available data their model has predicted the best route and also the duration that will be taken to travel from Sacramento to Floren. Now let us focus on prescriptive analytics. Prescriptive analytics describes the solution to a particular problem. What was the problem in this case on the predicted best route? Some predictions on the traffic congestions and that's when Google map recommends the substitute routes. Correct? Now these substitute routes are also prescribed by Google map as a recommendation. So prescriptive analytics is nothing but a solution and a recommendation provided for a problem. So in this case we have the best truth and we also have other substitute roots. So let's quickly summarize. Descriptive analytics is about summarizing the past data or to see what is happening for example in the Google map scenario. Predictive analytics is about what would happen and prescriptive analytics is about prescribing the solution, the best solution and the recommended solutions. Now let us refer back to Mark's earlier scenario. Mark along with his team identified the best model. They also did some testing and they got identified the best results and they also finalized on one particular model based on that particular model. Now the results have to be provided in such a way that they are the best results and also the recommendations. Correct? So it may not be just one single solution. it may be a solution with couple of other recommendations as well. So that is exactly what happens in the entire process of data analytics. I hope this has been clear so far. Yes. Okay. And now let us focus on benefits of using R. Why do companies extensively use R for data analysis and why is it chosen? Firstly, R is an open-source programming language which means that there is no license required to work with R and R does not require you to have a coding experience which means that a non-technical person in your team can also learn R very easily and start coding or building models in few lines of codes. R can also be used with other programming languages such as Java, C++ and Pythons. And the integration of R with other programming tools or BI tools is very simple and easy. And various statistical models are readily available in R. Which also means that there are plenty of inbuilt libraries and packages already available. And reporting the results of an analysis becomes easier by using these inbuilt packages and for creation of these models in just simple few lines code. With this understanding of the benefits of using R, let us quickly hop on to R studio and start performing the hands-on exercise for data analysis. For this exercise, we will use a data set named as demographics which is in a CSV file type. Firstly, let us load the data set to R studio and we will locate this in an variable named as demo. We also refer to this as a data frame. And now you will notice that a variable is created in an environment section which is in the bottom right hand side of the R studio window. And this particular variable comprises of 510 observations or records with eight variables. Let us simply expand this particular data frame and have a quick check on the data structure and understand the data types. This particular data frame includes variables such as age, marital, income. The unit of income is dollar per day, education levels, the car price, car category with several levels, gender and retired status. Now let us view the top six records of this particular data set. For this, let's simply type head of demo. And the result is now visible in the console section. If you're interested to view all the records, then simply type view of demo. And this new window will show you every single record that is being loaded to our studio. You may also simply apply the filters and the filter section here on individual categorical variables. Now that we have loaded the data set and also viewed individual records, let us focus on creating subsets of records by applying filters on individual variables or multiple variables. So firstly let us apply filter on gender. We will only retrieve the records with gender is equal to female. And we will locate these records in a variable named as demo 2. As you notice now in the environment section the second variable is also created that is demo and this now is comprising of 250 observation which means the records are filtered down to only gender female. Next let us see how to apply a filter on income variable. Let us only retrieve the records where income is greater than 100. Let's view the result. As we see here, all the records include income greater than 100. Now, let us modify this query and we will ensure that the retrieved records includes income greater than 100 and also specific variables are returned. Let's say we only want to have the first variable, third variable, and the seventh variable returned as the result. Let's have a quick check. So, we only have the first, third, and the seventh variable returned. How about we only exclude the variable 6 to 8? For this we include a prefix of minus sign. And now let us see what is the result. We have the variables from first to the fifth variable. However, we don't have six, seventh and eighth variable. I hope it's clear so far. Yes. All right. Now let us see how can we apply condition by including both the variables that is gender and income and then we will filter the record and create a subset of data. Now let's view the result. Income is greater than 100 and the gender is only female. This is one way of creating subsets. However, now let us see how to use the subset command and create the subset. Let's create a subset of records by applying filter on marital status and age. We'll only retrieve records where marital status is equal to married and age is greater than 35. Let's now view the result. So here we have the age greater than 35 and marital status is married. Let's use the same code and this time we will retrieve selected variables. Let's say variables ranging from 1 to 3. Let's have a quick check. So there are three variables. age is greater than 35 and marital status is married. Now let us see how to structure the data by sorting the data frame in ascending and in descending order. We will apply this order function on the variable income. Firstly let us see how to order income variable in ascending order. Let's do a quick check. And here we have income in ascending order. Now let's see how to modify the same code and view the records with income in descending order. So we have now the income in descending order. How about include two variables and sort the variables accordingly. Firstly, we will sort the records by ordering income and age in ascending order. Let's quickly view the result. We have income in ascending and age as well in ascending. Let's now modify this code. This time we will order income in descending and age in ascending. Let us view the result. So the income is in descending and age is in ascending order. So hope this is clear on how to solve the data frame by ascending and descending order per variable or by using multiple variables. With this we will focus on learning statistical analysis. How to perform statistical analysis on individual variable or multiple variable? Let's start by understanding the data distribution of variable income so that we identify what is the minimum value of income, what is the maximum, what is the range, what is median, what is the mean and we will also focus on the quantile distribution which is also analyzed in a box plot. What is the minimum value in the variable income? It's nine. And what is the maximum? So we have the maximum value. Now let us see what's the range. So the range shows you the result with minimum and the maximum value. How about the difference of maximum and the minimum? Now let us focus on other summaries of data distribution for this variable income. Let's identify what is the mean value of income. The mean is 78. Let's also understand what is the standard deviation. All right. So the standard deviation is $112. Let us see what is the variance. The variance should be larger than the standard deviation. Now let's say what is the median absolute deviation. As you notice here the median absolute deviation value is lower than standard deviation. Why do we make this comparison? From this it is evident that median absolute deviation is robust to outliers and standard deviation is sensitive to outliers and also to the change in the mean value. Now let us understand the quantile distribution. This is the same analysis that is visualized in a box plot ranging from 0% to 100% identifying the individual data points. And we can also refer and compare this to the min, the max and the median values. Let us quickly see what is the median value of income. As you notice here the median is 45 as well the 50th% of quantile is 45 which means 0% is minimum and 100% is the maximum value. Now if your question is what is 25% and 75%. This is again used for identifying the range of interquartile. The interquartile range is nothing but the difference of 75% minus the 25%. Let's quickly see what is the IQR of income. The IQR of income is 58. Let us do a quick check. 75% of quantile is 86 and 25% of quantile is 28 and the value is 58 which is equal to the IQR result. Now that we have focused on the statistical analysis of the individual variables data distribution, let us focus on the data visualization. In this we will have a pictorial representation of analysis to identify the outliers to see what is the minimum and where do we see the data densely populated and how is it scattered etc. We will begin with creating a histogram. Now histogram can be used for univaried analysis. Which means in this scenario we will consider income variable and we will see how the count of income ranges gets distributed in a histogram. For this we will have to install a package called as ggplot 2 and also call this library ggplot. Let us install the package. And now let us call the library. All right. And we are ready now to begin with visualization. For this we will use the geometric object histogram on the data demo data frame. Let me expand this window so that the code is visible and also use an aesthetic mapping for variable income. This will be helpful for filling colors or filtrations etc. and only include 30 bins with individual bin size width of 100 which means there will be 100 incomes in individual bins. Let's quickly look at the distribution of this histogram. As you notice there are couple of outliers. The counts of these income range are very limited. However, we see the densely populated income ranges with higher counts between 0 to $200 per day. This is also a way to identify and segment the customers based on their income ranges. Now, let us see how to change the color of this histogram and also the border of the histogram. For this we will include some additional options such as fill fill with blue color and the border color is black. Now as you notice here the executed code provides us the histogram with blue color bars and black color border lines. Now we will focus on creating a facet grid. Facet grid is also an aesthetic mapping object. We will see how to enable the multiple histograms across the marital status and the genders so that we identify how the income is distributed for individual marital status as well the genders. Let's zoom this view and have a look at it. As you notice here, there are some interesting outliers here in the data distribution. Female unmarried drawing higher income and male unmarried and married also drawing higher income as compared to the females. Whereas if you notice that the female unmarried is drawing much higher income than the male. This may also be very much related to the age. Now let us see how to create a stacked histogram. When I say a stacked histogram, I mean instead of filling the color, we will fill the gender. So that there is a stack within the histogram. So as you notice here, I have made couple of changes. I have included fill equal to gender within the aesthetic mapping. Now let us look at this histogram. As you see here the gender is filled in the histogram. Hence we have stacked distribution of female and the male. Now let us focus on creating a bar chart with education versus income where we can identify the education levels and the income ranges for these education levels. As you notice here, we are going to create a visualization where we have the aggregation in form of mean and the geometric object used here is bar plot. Now let's zoom this view and understand which education level have higher average income. So as we see here the blue color bar is the post undergraduate degree which means this education level draws higher average income as compared to other education levels. Now let's create a histogram where we will see car price and the number of cars for individual category. Let's look at this visualization. This visualization provides us some interesting insight just by looking at the distribution of the car prices and the counts of the cars at uh the car category economy and even the luxury. Luxury car category or car price is pretty much distributed. Whereas economy car category is densed which means that we could also look back into the income and age variables and try to figure out further more insights and then segment the customers for further targeting of these customers. Now what happens if we simply change this bin width to 30? As you observe here, changing the bin width or increasing the bin width will also reduce the number of bins. Now we only have four bins here and the car category is filled. That is what we have enabled within the aesthetic mapping. And we see some more interesting inside. As you look at the standard and the luxury car category, the car prices are pretty much overlapping for the car category, luxury and standard. This could be the starting car price of the luxury brands. Now let us create a clustered bar chart. Let's look at this visualization. In this visualization, as you observe, though we have enabled fill equal to gender in the aesthetic mapping, we do not have the view in stack form, but we have the bars one besides the other. It is also because we have enabled a position called as position equal to dodge in the code. Now what is the insight that we can draw from this visualization? As you see post-graduate degree with female gender is drawing higher average income as compared to any other education level. Now let us see how to create a box plot for variable income across the genders. So the box plot can be enabled if there is a biariate analysis to be performed on a continuous variable and a categorical variable or multiple categorical variables with a continuous variable. Now let's look at this visualization. What does this say? We have data distribution of income for individual genders that is for female and the male. And we also notice outliers here. Anything above this whisker is considered to be outliers. It might make more sense if we also include some coloring for these outliers. Maybe also enable shape. Now we have colored the outliers and it's colored orange. Let's see if we can also enable the shapes. And now we have here the outlier color as well the shape enabled. Now let us see how to enable a violin plot. What is the utility of violin plot? With a box plot we understand the analysis and the distribution of the data points is to identify the outliers to know what is the minimum value, what is the max, what is the median and what are the outliers. But what is the purpose of a violin plot? Let us have a quick check. as you observe there is some concentration of data points in the bottom of every car category. However, the concentration is higher for standard car category as compared to economy and the luxury. Now, this is an interesting insight that you wouldn't have come across in box plot. The box plot is a very good representation for identifying outliers. However, violin plot will help you focus on the nuances which is not captured by the box plot. We can also simply combine the box plot and the violin plot together. Simply include this jaw object. Let's zoom this. Now you have a representation of box plot and the violin plot both combined in a single visualization. Interestingly you notice the outliers as well the concentration in the bottom of this violin plot. So this could be some interesting insights that you draw and focus on these data points and understand what exactly is happening there. Now let's focus on the density plot that is density estimate of the histograms rather than just viewing the frequencies. Now we see the frequency in the y-axis across the income distributions. How about enabling the probability as true so that we enable the density instead of the frequency. So now we have the density in the y-axis and in the x-axis we still have the income distribution. This is the way of also adding a line plot which is a density plot on the histogram. Now as you observe here the density plot is not in the same level as the bar. So let us adjust this line. For this we will include adjust let's say equal to three. And now let us see how the visualization appears. Now the density plot is on the same level as the bar. Now let us see how to create a cross table for car category and gender. For this let us call the library dcr. Now let us create the visualization enabling cross table for car category and gender. Let's look at the result in console. As you see here now we see the counts of the gender for individual car category. The values over here represents that there are 67 females falling within the card category economy and 80 males within the card category economy. And for luxury we see that the count of female is higher than the male as well the proportions. Now how do you understand what proportions are presented here? We may simply turn off some of the proportions like the t test, the kai square etc. Let us see how to enable that. Now let's look at the result. This looks better now that we have the counts, the female counts and the male counts across individual car category. We also see the percentages rather than just looking at the absolute value. So there are 45.6% of female within the car category economy and 54.4%age of male within the car category economy. Similarly across rest of the car categories. This kind of cross table or a contingency table is also helpful when you want to analyze the different categorical variables and identify the counts or the proportions. Now let us see how to use a scatter plot of age versus income. Scatter plot is a visualization used for bariate analysis. When you want to perform some analysis between two continuous variable at a data point level rather than performing the analysis at an aggregated level such as sum or mean. And now we have a scatter plot of age versus the income. Age in the x-axis and income in the y-axis. Though we do not see any kind of a positive correlation or a negative correlation, but we still see some interesting insights over here. Some of the data points are pretty much scattered and much away from densely populated data points. I hope the learning has been informative and interesting. So far we have covered the concepts of data analytics as well we have performed some hands-on doing some statistical analysis and also creating interesting visualization to learn more and further on data analysis with R. Please watch the video above. >> Everyone so welcome to this session where we will learn on time series analysis using R programming language. So this is basically a mini project where we will look at time series data and how we can analyze it, visualize it to basically find some important information or gather insights from the data. Now when you talk about time series analysis, time series is basically any data set where your values are measured at different points in time. So when you talk about time series data data is usually uniformly spaced at a specific frequency. For example, hourly weather measurements, you have daily counts of website visits, monthly sales total and so on. So when you talk about time series that can also be irregularly spaced and sporadic. For example, time stamp data in computer systems event log or history of 911 emergency calls. Now when we work with time series data for example here I'm taking a energy data set we can see how techniques such as timebased indexing resampling rolling windows can help us explore variations in electricity demand and renewable energy supply over time. Now here we will look at some aspects of this data set which I am considering. So there is this is open power systems data set and here is the data set I have. We can look at the data set. Now this is in a simple format. It has time. It basically has values for consumption and then you have data for wind and solar and wind plus solar. So in certain cases you have only the date and the consumption. But then if we scroll down we will also find data for wind. solar, wind plus solar and so on. So this is a time series data set which we would want to work on. Sometimes you may also have the data collected which just does not have the time but it may also have time stamp that is it would have say hour, minutes and seconds and that can also be worked upon. So let's consider this data set and let's work on this project where we will analyze this time series data set. Now here we can work on this time series data. We can basically create some data structures out of it such as data frames. We can do some time based indexing. We can visualize the data. We can look at the seasonality in the data. Look at some frequencies and also do some trend detection. Now when you talk about this data set, it has electricity production and consumption which is reported as daily totals in gawatt hours and here are the columns of the data which I was just showing you. So you have data, you have consumption, you have wind, you have solar and wind plus solar. So this is the data we have and we will basically explore say electricity consumption and production in Germany which has varied over time. So some of the questions which we can answer here is when is electricity consumption typically highest and lowest? How do wind and solar power production vary with seasons of the year? What are the long-term trends in electricity consumption, solar power and wind power? How do wind and solar power production compare with electricity consumption and how has this ratio changed over time? We can also do wrangling or cleaning of this data or pre-processing of data and create a data frame and then we can visualize this. Now let's see how do we do that. So I will open up my R studio and let's look at the data set. So here is the data set. Now I'm picking it up from my machine. You can also pick it up from GitHub. So all the data sets or similar data sets can be find in my GitHub repository. And here I can look in the data sets. You will find lot of different data sets here. There are some time series data sets such as power. I can search for power or you have basically coal or you have this opsd Germany daily data set and there are many other data sets which you can work on. Now to get the documentation on this project, you can also look in my GitHub repository and you can search for repositories and then basically you can look in data science and R and here there is a project folder where I have given the documentation sample data set and also your time series analysis related document. This is also the code which you can directly import in your R studio and you can practice or work on this project. So let's see how does that work. So first thing is we will create a data frame from this data set. Now here if you see I am using header as true so that it understands the heading of each column. I'm also giving row names and I'm specifying date. So there is this date column in the data set as I showed you earlier. Let's look at it again. So you have date, consumption, wind, solar, wind plus solar. So you can suggest that date should become the index column which can be useful. So you can do this. Now let's just create this. Let's look at what does this data frame contain. And here if you see it shows me some data which has been now as a part of this data frame structure it starts with consumption wind solar wind plus solar and if you see this one is becoming my index column. So I can always do a head and look at part of the data frame using head or tail. So look at the first records. So let's see this now that shows me the head data. I can also do a tail and look at the ending values. So if you closely see here we have wind solar wind dot solar and that basically has NA values. So there are missing values but let's look at the tail and that tells me that there is some data available for wind and solar and wind solar. Now we can always look in a tabular format using view and we can look at the data. So this shows me that there are values in these columns. We see NA values but if I really scroll down I can see some values which would be available for wind and solar and wind solar. So I can just use view. Now I can look at the dimensions of this particular object and that tells me there are 400 4,384 rows and four columns. You can always look at the structure that is check the data type of each column which can be very useful. So if I see here I don't see the date column because date column was considered as an index which can be useful. But I also look at my other columns. So they are of the num types. So that's the data type for each attribute or each column here. Now we would be interested in looking at this date column. So let's look at the data type of this date column. Now if I try to do this, this will show me that this is null because date as a column does not exist because we created it as an index. So if I look at row names and then I search for my data show me the index column or row dot names it tells me these are the values that's the date column which we are seeing here. Now we can access a specific row by just doing a my data and give the index value or row name value. So let's look at that and that shows me based on this index you're looking at the value. You can obviously search for a different date something like this. You can also pass in a vector and you can give range of values. So that is 0 1 2006 to 4 of January. And we can look at this one. So it shows me these are the values. So here actually I'm not giving a range but I'm just selecting multiple values from row.names. Now we already know that in R you have a summary function. So you can always do a summary and that gives you for each column it gives you minimum, first quartile, median, mean, third quartile and maximum values. So we are looking at consumption, we are looking at wind, solar and wind dots solar. Now this is good. But then if I would want to really visualize the data, access the data, do some analysis, then it would be good to take all the columns and then we can later decide to change the data type of say date column if we want to use it. So earlier I was using date as row.names or the name of the rows or index what you call in any other programming language. So here I will just use my data set and I'll say header as true. I'm calling it my data 2. Let's look at the data. And this one shows me five columns wherein my first column is the date, consumption, wind, solar and so on. Now looking at the structure. So let's look at the data type. So it tells me that if now I'm interested in looking at the date column from my data to data frame, it tells me it is a factor with 4 384 levels and these are the values. So it is not in a datetime form