[UTokyo Matsuo Lab GCI Report] Sessions 2-5 | What is the ultimate purpose of data analysis?

Introduction

Hello.

I am Yuika Tsukamoto, a student at Kamiyama Marugoto College of Design and a sales intern at EQUES Inc.

In my previous course report, I introduced the overall overview of GCI and how to proceed with the learning.

This time, I will summarize and reflect on the content from the second to the fifth session.

From the second to the fifth session, the content mainly focused on the basics using Python. We covered everything from how to use Google Colab to basic Python syntax, NumPy, Pandas, Matplotlib, and finally, simple linear regression.

At first, I had a strong impression that these were 'Python learning sessions,' but after taking them over several sessions, I felt that I was learning not just how to write code, but the process of handling, organizing, and visualizing data to think from it.

Starting by running Python

In the second session, we learned how to use Google Colab and the basic syntax of Python.

Google Colab is a tool that allows you to execute Python in a browser. During the lecture, the explanation proceeded while showing the actual operation of Google Colab, so it was easy to understand where we were and follow the content.

In this session, we covered basic Python syntax such as arithmetic operations (addition, subtraction, multiplication, division), conditional branching, and functions.

If you complete the pre-study, you can follow the lecture content relatively smoothly. Conversely, I felt that if I took this without looking at anything beforehand, it would be a bit difficult in the second half.

Turning data into a 'calculable format'

In the third session, we learned how to use NumPy.

NumPy is a Python library suitable for scientific computing. It allows for the calculation of N-dimensional arrays and handling data like tables in bulk.

The lecture mainly covered up to two-dimensional arrays.

What left an impression on me here was that NumPy was used as a tool to interpret tables.

It was explained that the purpose is not just to perform calculations, but to process information and make it ready for analysis.

It was also suggested that there is no need to memorize everything, and that it is fine to learn by looking things up as needed.

Since libraries have many functions, I felt it is more important to first get a rough idea of what they can do rather than trying to memorize everything from the start.

Making data "easy to handle"

In the fourth session, we learned how to use Pandas.

Pandas is a library that allows for the processing and manipulation of data in DataFrame format. It was explained that it is frequently used in data analysis.

For the basics this time, we learned how to extract desired information from data, how to split data, and how to handle missing or anomalous values.

I felt that the content started to become a bit more difficult around this point.

However, by seeing interactions related to the homework on Slack and other platforms, I was often able to notice my own mistakes.

There are parts I might overlook when working alone, but by seeing other people's questions and answers, I can sometimes realize, "Oh, I might be making the same mistake here."

Making data "visible"

In the fifth session, we learned how to use Matplotlib.

Matplotlib is a library for visualizing data using programs.

By changing functions, you can create various graphs such as scatter plots, bar charts, pie charts, and box plots. You can also change font sizes, graph sizes, and more.

Information that is hard to understand as raw numbers becomes much easier to see when visualized as a graph.

On the other hand, because there are so many types of functions, I felt that the amount of things to memorize increased all at once here.

Since the graphs and functions used depend on what you want to see, it seems necessary not just to memorize the syntax, but to use them selectively according to your purpose.

Using data to make predictions

In the fifth session, we also learned about simple linear regression analysis.

Honestly, I found it quite difficult.

Simple linear regression analysis involves drawing a straight line that approximates the relationship between two sets of data and using that to make predictions.

In the lecture, an example was given of predicting third-semester grades based on first-semester grades.

In short, I understood it as using past data to anticipate what might happen in the future.

We were told that we would use something called a simple linear regression model to make predictions, but I cannot say that I have fully grasped this part yet.

However, looking back at the flow so far, I can see a connection.

Learning to handle data with Python.

Processing data with NumPy and Pandas.

Visualizing data with Matplotlib.

And then, connecting it to predictions with simple linear regression analysis.

I felt that data analysis is not just about looking at numbers, but about putting data into a usable format, making it easy to see, and using it to make decisions and predictions.

Thoughts on working on the assignments

As the sessions progressed, there have been more instances in the homework where I need to refer back to previous content.

Therefore, if I don't remember the content, even vaguely, it can be difficult to figure out where to start.

Personally, when I think I might be able to recall previous material, I often ask GPT to confirm the method, and if I still don't understand, I review the slides.

As new content continues to be added, I felt that even if it is not necessary to remember past content perfectly, it is important to have a rough idea of 'what was learned and where'.

Conclusion

Throughout sessions 2 through 5, I learned everything from the basics of Python to data processing, visualization, and the fundamentals of prediction.

At first, these seemed like separate topics, but looking back,

・Handling data
・Organizing data
・Visualizing data
・Predicting from data

I feel that it followed this flow.

To the question, "What is the ultimate purpose of data analysis?"

At this point, I believe it is the preparation and practice required to make decisions and predictions based on data.

There are still many difficult parts, but I want to continue deepening my understanding little by little through lectures and assignments!

Recruitment | Would you like to become a "Partner" for a startup from UTokyo's Matsuo Lab?

The official EQUES Inc. LINE OpenChat provides a wide range of information, including one-off part-time jobs, internships, recruitment information, the latest AI news, and networking event details✉️
If you are interested (anonymous participation is of course welcome!), please join us via the link below🌟

🍣 Announcement of Networking Event 🍕

We hold a "Regular Networking Event" at our head office on the third Friday of every month!🌟
・You have even the slightest interest in AI
・You want to talk about AI trends or life at UTokyo over sushi and pizza
If this sounds like you, please feel free to join us. Participation is free, and you are welcome to come and go as you please🌟

Click here for the participation application form:https://docs.google.com/forms/d/e/1FAIpQLSfNUpZ9ag1cUERqJO2V5z9QcOoNtiRBlqG4MKajlwiiOqlnHw/viewform

#AI　#HighSchoolStudent　#Data　#AISideHustle　#Entrepreneur　#Internship　#UTokyo　#DataScience　#Lecture　#KamiyamaMarugotoKosen　#UTokyoMatsuoLab　#GCI　#Python