Photo by the author
# Entry
For this article, I selected five projects that cover different stages of a typical data science workflow, from basic data cleaning to data mining, model building, and even deploying them for real-world exploit.
# 1. The ONLY data cleansing environment you need
This movie is by Christine Jiang, who works as a data scientist and shares with us a really practical approach to data cleaning that I think will be useful for anyone working on projects. When cleaning data, we often wonder ‘how spotless is enough’, and Christine shows you a clear way to go about it using her five-step CLEAN framework. It shows you how to find solvable and unsolvable problems, align values, document everything, and iterate to ensure data is reliable without striving for “perfection.” The examples he uses, such as fixing missing country codes or inconsistent product descriptions, are very accessible, and the thinking he highlights is just as essential as the tools. I believe this is an extremely practical guide for anyone trying to exploit real-world data effectively.
# 2. Exploratory data analysis in Pandas
This movie shows why simply having data is not enough and how careful analysis of the numbers can reveal hidden patterns. The presenter walks you through examining datasets, summarizing distributions, checking for missing values and outliers, and visualizing relationships between columns using pandas AND Seaborn. I find this really practical because it not only shows the commands, but explains why each step is essential and how statistics can tell you things that are not obvious at first glance. This is a great guide for anyone who wants to explore real-world data and obtain relevant information before modeling.
# 3. Data visualization using Pandas and Plotly
This movie by Greg Kamadt, founder of Data Independent, shows how telling stories with data is as essential as building models. Walks you through a hands-on tutorial using pandas for falsifying data and Plot in the case of interactive charts, starting with the basics of visualization effectiveness. You’ll see how to load and shape data, select the right chart types, and add formatting changes to make your charts clear and uncomplicated to understand. I really liked that it was practical, with tips on how to solve real-world problems like outliers, date lines, and aggregations, as well as how diminutive choices can improve readability. By the end, you’ll know how to create interactive, shareable charts that effectively communicate insights.
# 4. Feature engineering techniques in machine learning in Python
Once the data is spotless and understandable, it’s time to create better features. This guide focuses on the “feature engineering” stage, where you transform and generate novel columns of data that can make your model smarter. The instructor explains techniques such as coding categorical variables, handling missing data, dimensionality reduction (principal component analysis (PCA)), and creating interaction terms. I like that it also highlights what not to do, such as data leakage, over-fitting, and over-designing features. This is a great resource for anyone looking to move from raw data to creating well-designed features for real-world machine learning.
# 5. Deploying the machine learning model in Streamlit and creating live forecasts
Finally, the most satisfying part – bringing the model to life. In this tutorialYiannis Pitsillides shows how to deploy a trained machine learning model using Streamlined. It loads a saved model, sets up a spotless interface with input fields and buttons, and generates real-time car price forecasts. The video even includes a visualization of the importance of the feature Plotlyso you can see which inputs are most essential. I liked that it was hands-on, with tips on separating raw and cleansed data, handling dependencies, and running applications locally or on a host. It’s a compact tutorial, but it does its job perfectly and provides a comprehensive experience that most beginners lack.
# Summary
These projects cover all key steps in the data science workflow and show how theory becomes reality in practice. Grab your datasets and start experimenting. There’s no better way to learn data science than by doing it.
Kanwal Mehreen is a machine learning engineer and technical writer with a deep passion for data science and the intersection of artificial intelligence and medicine. She is co-author of the e-book “Maximizing Productivity with ChatGPT”. As a 2022 Google Generation Scholar for APAC, she promotes diversity and academic excellence. She is also recognized as a Teradata Diversity in Tech Scholar, a Mitacs Globalink Research Scholar, and a Harvard WeCode Scholar. Kanwal is a staunch supporter of change and founded FEMCodes to empower women in STEM fields.