Data Analytics for beginners like me

Tram Ho

Prologue

This time, for some reason, I am studying Data Science, so the articles in this time will revolve around this topic. In online courses, I’m in the middle of a “Learn data analytics for beginners” course by SkillUp . This course in my opinion is relatively good, giving an overview of Data analytics. In this article, I will take notes to discuss with you some of the first content of this course (you can see Part 1 in the subtitle of my article).

Data analytics is data analysis, what else?

  • DA is the process of examining and analyzing raw data to draw graphs, find information from the data, and thereby improve the quality of business, products, and services. It is used not only in business but also in science and research to test scientific models and theories.
  • In the simplest terms, it helps you analyze the data you have: extracting useful information with science.
  • Data analytics (DA) plays an important role in making more scientific decisions, helping businesses operate more efficiently with multi-way, multi-dimensional and technical diversity.
  • My peers when I was still a student, surely every time I heard of accounting, I would see images of people recording data on paper through movies, looking very cumbersome and difficult. boring, but sometimes a sleepy man fills in a few more zeros and goes, and sometimes can’t find it. Thanks to the development of technology, calculation tools like Excel gradually dominate, especially in the accounting industry, but this method is considered easy but does not fully solve the problem. Cash flow management is very important in businesses, banks or Startups, while the traditional way is difficult to track, leading to difficult control.
  • According to the advice of the teacher of this course, to be able to improve your project skills and make your decisions easily, you need to follow the flow of the project process in your data analysis process as follows:
    • Define your goal (goal)
    • Identify the right metrics to measure
    • Collect and extract data from multiple sources
    • Explore and analyze data
    • Interpret and visualize data
    • Infer from data to make decisions

Data analytics lifecycle

  • Discovery: That is, mining information about business subjects and assessing the feasibility of existing resources to meet the goal or not.
  • Data preparation: Perform ELT (Extract, Load and Transform) process for data
  • Model planning: Define techniques and data to plan to build a model with the goal of understanding the correlation between variables in the data set
  • Model building: Build and develop data analytics for testing, training and production
  • Communicate Results: Identify important findings, valuable business value,
  • Operationalize: Handing over Final reports, code and technical documents

Data analytics is also divided into several types

There are 4 types (techniques) of Data Analytics, and they combined can provide answers to everything a business needs.

Descriptive Analytics: What happened?

  • This technique brings out a lot of information from the past: who, when, where, etc.
  • Focus on describing an overview of facts.
  • The goal is to summarize the information found and understand what it is doing (in the past).
  • There are 2 sub-technologies: Data aggregation and Data mining
    • Data aggregation: Information aggregation technique, some tools such as MS EXcel, Matlab, …
    • Data mining: The process of classifying and organizing large data sets to identify patterns and establish relationships to solve problems through data analysis.

Diagnostic Analytics: Why did this happen?

  • This technique focuses on past data to answer the question of why it happened in the past
  • A deep look at the root of the problem to understand the cause of the event but with limited possibilities
  • Only provides understanding of normal (i.e. not very deep) relationships and linkages in retrospect.
  • Some techniques can be followed like drill down, data discovery, data mining and correlations

Predictive Analytics: What will happen?

  • This technique predicts the probability of an event happening
  • Built upon the preliminary analysis description phase to derive the probabilities of the outcomes.
  • For example, when you post something on Facebook or Instagram, this technique will analyze whether your interest or sentiment on a topic is positive, negative or neutral (customer behavior prediction problem).
  • Some typical Machine Learning models in this technique such as Random forest, SVM, …

Prescriptive Analytics: How can we make it happen?

  • Provide solutions for future predictions
  • It creates and updates relationships between actions and results using powerful feedback systems.
  • Not only does it aid in suggestion optimization in the decision-making process, it also helps to reduce the risk of seeing dependencies between available analytic predictions.

You may think that you need to follow the above techniques sequentially, but it is not required to use all 4 techniques sequentially, even most companies jump straight to Prescriptive Analytics. It is a nascent stage of implementation and not many firms have used its full power yet. However with advances in predictive analytics will certainly pave the way for its growth.

Analyze a small problem of the big Amazon to better understand the above boring theories

4 The above technique sounds too theoretical, let’s analyze an example below to understand better.

  • Using Diagnostic analytics , Amazon knows that in the last year, revenue in the West Coast increased; Possibly the reason is increased spending on sales training for sales managers
  • Using Predictive Analytics , Amazon analyzes purchase history to consider various factors such as price, time, weather, festival times, etc.; Based on that, it is possible to predict that next year West Coast’s revenue will increase by 10-12%, but how to achieve that? (–> Descriptive analysis)
  • Using Description Analysis , Amazon discovered that it costs about $ 20 million to invest in different sales training courses
  • You can go straight to Prescriptive analytics to find out which training courses have a good return on investment (ROI) and implement an optimization plan: which training programs will be dropped, the program What training is continued to maximize profits

A few other small content

  • Some great benefits of Data Analytics:
    • Help define target customers based on some information such as where customers buy, brands or products that users search for the most.
    • Using data for e-commerce, you can manage investments and predict demand by determining what time of year customers shop the most. You can rely on emotions or customer feedback about the price to optimize the price for the product
  • Some Data Analytics tools: Power BI, Tableau, Logi, … In which my upcoming goal will be to learn about Power BI, please read along ^^

Epilogue

My recent posts are full of words, if you can read this far, I hope you know that I am very happy. The content of the article is the ideas I have distilled from the course as well as explaining it again to make it easier to understand (English sub so many things are a bit difficult to understand), so I also hope to receive comments or suggestions from everybody.

Share the news now

Source : Viblo