Activity when ML is not applicable

[yuliaUU]

this can be done as activity: giving each team dataset, its description and research question. and ask whether it is ok to use ML:

[yuliaUU]

Sonar Dataset • Description: This dataset involves predicting whether an object is a mine or a rock based on sonar returns. It includes 208 observations with 60 input variables. • Link: Connectionist Bench (Sonar, Mines vs. Rocks) - UCI Machine Learning Repository • Question: Can we accurately classify sonar returns as either mines or rocks? • Issues with Dataset: The small size (208 instances and 60 features) and imbalanced class distribution make it difficult for machine learning models to generalize well, often resulting in overfitting. The limited number of observations may not be sufficient to train a robust model effectively.

[yuliaUU]

Iris Dataset • Description: The Iris dataset contains 150 instances of iris flowers, each described by four features (sepal length, sepal width, petal length, and petal width). It's one of the most famous datasets in the machine learning community. • Link: UCI Machine Learning Repository - Iris Dataset • Question: Can we classify the species of an iris flower based on its features? • Issues with Dataset: The dataset is too simplistic and small to be representative of more complex real-world problems. It's often used for educational purposes but doesn't provide sufficient complexity for practical machine learning challenges.

[yuliaUU]

This on is tricky: cause you can use ML- but tehre is a "but" Weather Forecasting Dataset

• Description: Weather forecasting datasets typically include atmospheric data such as temperature, humidity, wind speed, and pressure over various time periods and locations. These datasets are generated using physical models that simulate atmospheric dynamics based on deterministic equations, such as the Navier-Stokes equations.

• Link: https://github.com/florian-huber/weather_prediction_dataset

• Question to Use ML For: Can we forecast teh weather based on teh environmental conditions in a given city?

• Issues with Dataset:

  • Deterministic Nature of Weather Systems: Traditional weather prediction models rely on physical laws and deterministic equations to model atmospheric conditions accurately. Machine learning models, while faster, lack an understanding of these
  • Machine learning can produce forecasts that are not physically realistic, such as negative humidity levels or temperatures outside expected ranges. This is problematic because accurate weather forecasting depends on precise adherence to physical laws.