Hello, this is Marketing Data Science Lab. It's now the year 2024, and I hope all your planned endeavors for the year are successful. As I upload this video, I see that many of you are interested in data analysis. If you want to start learning data analysis but don't know where to begin, it can be quite daunting. So, what's the best assistant and no-code tool? I've planned to use ChatGPT to study the fundamentals of data analysis. I've created 50 lessons, and if you study 1 to 2 lessons per day, you can study 5 to 10 lessons a day. Within a few days, you'll be able to study the fundamental theory needed for data analysis. Even if you're not a data scientist or developer, understanding how data is analyzed and extracting meaningful insights can be valuable for anyone. Moreover, with the abundance of no-code tools available, if you understand the basic theory of data analysis, you can visualize data and even try your hand at data analysis modeling. And if needed, you can gradually progress toward learning Python and its libraries. Now, before creating GPT, let's work on the lecture topics that will be used as input to GPT. Although the content may vary slightly as each chapter is generated, I want to plan and provide the curriculum to GPT using ChatGPT. (I've provided the following prompt to ChatGPT) You are an excellent instructor for data analysis, and I'd like you to create a curriculum for people who know nothing about data analysis. Also, create a curriculum on how to start data analysis using no-code tools like code interpreters. The curriculum consists of 50 lessons, aimed at those unfamiliar with data analysis. It should cover the basics, data analysis process, introduce data types, and explore data analysis using statistics, machine learning, and more. The goal is to understand how to perform data analysis up to a practical level using no-code tools. So, I asked for 50 lecture titles, and ChatGPT provided them well. These lectures are divided into three main branches: Data Analysis Definitions, Theory, and Process. Let's go through the chapters one by one. The first chapter covers the introduction to data analysis, and the next, role of data in decision-making, types of data, and an introduction to no-code data analysis tools. It also includes an overview of the data analysis process, including data collection, cleansing, and visualization. Next, moves on to visualization techniques, how to create effective charts and graphs, and an introduction to no-code tools for data visualization. With the introductory and data analysis process sections complete, then into the world of statistics, starting with an introduction to descriptive statistics, understanding measures of central tendency, variance, and more. Then, using no-code tools to explore data in terms of descriptive statistics. Once descriptive statistics are covered, moves on to inferential statistics, introducing hypothesis testing, correlation vs. causation, and regression analysis. After covering time series analysis and concluding statistical analysis, moves on to machine learning, beginning with the basics of machine learning, supervised vs. unsupervised learning. Next, moves on to classification, and upon completing these topics, moves on to deep learning. Once deep learning is covered, case studies for each domain, effectively communicating data analysis results, data analysis project management, collaboration in teams, career paths, and more. However, for those curious about the overall data analysis process, before jumping into data collection at 4, it would be beneficial to provide an overview of the entire data analysis process. Therefore, after 4, before diving into data collection, I added a section that gives an overview of the data analysis process as a whole, covering data cleansing, analysis, evaluation, and more. I requested this addition by removing Review Session 49, resulting in a completed list of 50 topics. Now, I will provide this list to GPT and set up its behavior. The first question GPT Builder always asks is what kind of service or GPT I want to create. For those who don't know much about data analysis and are unsure where to start, I expressed my desire to create a GPT that can quickly teach the basics of data analysis, and ChatGPT suggested a name, 'Data Analysis Guide' And like the lecture topics, I entered the name "Data Zero to Hero" as I prepared with ChatGPT already. As a result, a profile image was generated. However, the image looked quite outdated. To create a clean text-based logo, I entered a prompt to design a logo and received a simplified version. The logo seems decent, and when asked if there's a specific focus within data analysis, I mentioned that it consists of 50 lectures aimed at those new to data analysis. The main topics include what data analysis is, the data analysis process, and methods of data analysis. So now that it's generated well, when user queries are confusing, should I request clarification explicitly or make educated guesses for the best answer? I chose to request clear questions for now. Finally, regarding GPT's tone, I intended to choose "Friendly," but it was already updated before I could select it, so it's set to "Friendly" now. Next, I'll go to "Configure" and modify the description to say "50 steps for beginners in data analysis." Then, I'll review the instructions generated by GPT. It mentions covering these key areas, balancing expertise and accessibility, and the goal is to create a process where beginners learn well and gain confidence. I will copy and input the lecture topics I organized earlier for GPT. There are 50 lectures covering these three main areas. I have entered the title here for GPT. Each lecture shouldn't be too long, around 1 to 2 minutes in length. Finally, I requested that users be asked if they understood and for questions or examples. The language used is Korean, so GPT will respond in Korean when users speak. I requested that it responds in English if I speak in English. Also, just like last time, I uploaded Kaggle's IBM Watson Marketing Customer Data to enable the experience of the data analysis process. And instead of just doing everything at once, I provided steps for data analysis. In the previous video, I tested creating a GPT,
and the content is the same as what came out there. Now, Conversation Starter always returns to check the 50-lecture list and start, so I added the content to show the 50-lecture list, and experience the data analysis process
talk about the first lecture, with the dataset I uploaded earlier. I used the same dataset downloaded from Kaggle for IBM Watson, as mentioned earlier (link). Now, let's move on to the test. So, I asked to show the list of 50 lectures, and it is generating a list of 50 lectures along with brief explanations. So, the introductory part of data analysis and the role of data in decision making are well included. Now, it will continue generating the remaining 50 lectures. I will always ask if you understand the lecture content and if you have any questions. So, I asked to talk about lecture 1 right away, and I requested it to respond in Korean if I speak in Korean. It responded well in Korean. So, number 1 is about the introduction to data analysis, what data analysis does, and how it is structured. As I write it, if you just read it through, you should be able to digest the content
within 1 minute, which isn't difficult. Now, I will ask if you understood the lecture content. During the analysis process, can you briefly explain the importance of data cleansing and preparation? You can answer. Then, when I asked to talk about the next lecture, Number 2, the importance of data in decision making, it started talking about how it provides objective evidence that can replace subjective judgments and assumptions. Various content like that is included in the front part, which is theoretical and relatively easy. Lastly, it asked how data can provide benefits in business, such as identifying opportunities, managing risks, and measuring performance. You can answer if you can explain how data can provide these benefits. So, instead of moving on to lecture 3 right away, I asked it to talk about lecture 4, which is about introducing the no-code data analysis tools. In no-code data analysis tools, various tools that allow data analysis without coding are introduced. So, you can visualize and analyze these functions with drag and drop. The main tools introduced are Tableau, Microsoft Power BI, and platforms like code interpreters. Tableau is mainly for data visualization, Power BI is a no-code tool for data analysis and sharing, and code interpreters can do both analysis and visualization. Now, I will ask questions. So, up until now, if you liked the content, you can copy it. You can copy it directly, or you can ask it to save as a Word document. So, I asked it to copy the lecture content and save it as a Word document. This way, it doesn't copy everything, but usually, it summarizes and saves it. So, it's more condensed than what I had in mind. If you really like the content, because even for the same chapter,
it's also good to copy and save it in Word, if you select "Show 50 chapters" and choose, the content may vary slightly each time it generates. So, creating your own notes by copying and pasting them into a program like Word is also a good idea. Because even without specifically asking for summaries, when the content is long, ChatGPT seems to work this way. Now, let's check the Word document. When I did, it had quite a bit of main content rather than a summary. It had the content that came out as mentioned earlier, briefly, a somewhat summarized version. So, I looked at the code that created the Word document, and there was no specific request for a summary. It created the Word document using the "docs" library, and attached titles and content like that. As you go further back, topics like statistics and machine learning are included. Even though the content is short, it can take a lot of time to understand, and you can ask many questions to ChatGPT. Especially for abstract concepts, I think it's faster to study the theory first and then think about it with easy examples or try implementing it with no-code tools. Up to now, I tried creating a GPT for a quick basic course on data analysis. GPT itself is free, but to use GPT, you need to be a ChatGPT Plus subscriber. Those who are already ChatGPT Plus subscribers are fine, but GPT-4, code interpreters, GPT, and Dalri3 are services that you can use if you subscribe to OpenAI's $20 per month subscription. In the video description, I'll put a link to the Data Hero GPT I created today, and if it helps you start your data analysis journey, I'd appreciate your feedback. I will update it whenever I have new ideas. Thank you.
