Data Analytics Types & Techniques

In this video, we'll explore and elaborate the various types of analytics that we can implement in our applications and businesses today. Data analytics is a higher abstraction of machine learning, and machine learning can be understood as a methodology of analytics. Now, all the applications may have different objectives in terms of analytics, and this can lead to various types of analytics that we can in turn use to get the appropriate outcome. There are four different types of analytics that we can apply to gain insight of the data and get a useful information that in turn can aid our decision making process. The first one is perspective analytics which focuses on the actions that we need to take in order to improve the existing outcomes. Now prescriptive analytics is considered a valuable type of analytics because we can use it to decide the future steps on the basic recommendations derived from the set of rules.

[Video description begins] Prescriptive analytics results in rules and recommendations for the next steps. [Video description ends]

Predictive analytics is another important type of analytics which provides certain scenarios to help predict the outcome of a given scenario. And predictive analytics plays an important role in various scenarios where we're analyzing historical data to facilitate predictive forecasting. The third important type of analytics is the diagnostic analytics, which plays a very important role in identifying the underlining factors of past performances and the results of the analysis can help set the right direction for the future.

[Video description begins] The result of the diagnostic analytics is often an analytic dashboard. [Video description ends]

In the current system, the diagnostic analytics is represented by analytical dashboards. Finally, the descriptive analytics provides focuses on the conventional approach of mining data to identify the current status space of the incoming data, and its impact on the outcome. Now let's explore an analytical scenario using the sample implementation to understand how analytics and its various derivatives are used or implemented in a business today. Every organization will have their own data stores to store their business data. The data store can either be business databases or cloud based storages. The businesses or organizations would need to apply descriptive analytics on the store data to understand what happened. This will help the business understand and explore further business opportunities.

Then, when they apply predictive analytics, they'll get to understand the current state. Basically, what's happening and why. And the outcome will help them predict the opportunities that could actually be profitable for the business. And finally when the business apply prescriptive analytics, they'll understand how to handle the current state and the predicted opportunities. So they can reach a different level and take the benefit of predicted opportunities. Now in the business applies, all the analytics it will help the business or organization increase their business, increase business performance, and provide values to other customers. Now let's explore how we can implement business analytics to help the decision making process using the sample implementation figure.

[Video description begins] A diagram appears on the screen. It shows the four steps of a business analytics decision-making process. The first step is Data source. This is a perception of disequilibrium. It involves observing and becoming aware of potential problem situations. The second step is Descriptive analytic analysis. It is a diagnostic process. It attempts to understand what is happening in a particular situation. The third step is Predictive analytic analysis. It is a problem statement. It involves identifying and stating problems and solution strategies in relation to organization goals and objectives. The last step is Prescriptive analytic analysis. This is a solution strategy selection. It involves selecting optimal course of action for the organization from the determined strategies. It also includes implementation of the strategies. The result of these steps would be a measurable increase in business value and performance. [Video description ends]

When we want to implement business analytics, there are certain steps, that we will follow or adopt to get a conclusive or desired outcome. We'll begin by identifying the potential problems. Following which, we'll implement a process of descriptive analysis to understand what's happening in the existing process. In the third step, we'll implement predictive analysis to identify the state of the problems, derive certain strategies related to the business goals and objectives, to resolve the problems and predict expected outcomes. And finally, we'll implement prescriptive analytics, which provides the capability of adopting the right strategy or course of action from among the various available or provided solution strategies that were identified during the predictive analysis. Now when we apply all these steps in business analytics, it will help us ensure the strategic decision making process is simpler and accurate for the strategic decision makers of the business and organizations.

Now Python provides various libraries that we can use to implement machine learning and analytics in applications. And Python probably is the most important language that one can adopt because its rich ecosystem, and the simplicity that it facilitates. Now let's explore some of the prominent Python libraries that we'll use to implement analytics. The first one is Pandas, and this library is extensively used for data manipulation and analysis. The second one is StatsModels, and this library provides rich modules that enable users to explore data and implement various models using the statistical approach. Scikit-learn is another Python library which is primarily used to implement supervised, unsupervised, and semi-supervised algorithms. It's open source and one of the most popular and commonly used Python libraries for implementing complicated algorithms like classification, regressions, and clustering. Mlpy is another Python library which is intended for machine learning, and this library is basically an extension of NumPy or SciPy. Finally, SciPy is a popular Python library, which is used for scientific and technical computing and it provides direct functions that we can use to simplify developer's tasks.