Hansstem / TDT4173-ML

Repo for machine learning TDT4173
0 stars 0 forks source link

Hvilke modeller vi skal lese mer på. (Hvilke kan passe) #4

Open Hansstem opened 1 year ago

Hansstem commented 1 year ago

Linear Regression: Use case: Predicting a continuous target variable based on one or more independent variables. Strengths: Simplicity, interpretability, and good for modeling linear relationships. Considerations: Assumes a linear relationship between variables, may not perform well with highly nonlinear data.

Random Forest: Use case: Classification or regression tasks, feature selection, and handling missing data. Strengths: High predictive accuracy, handles complex interactions, robust to outliers. Considerations: May overfit with noisy data, can be computationally expensive for large datasets.

Gradient Boosting (e.g., XGBoost, LightGBM): Use case: Both classification and regression problems, often used for structured/tabular data. Strengths: Excellent predictive power, handles complex relationships, robust to outliers. Considerations: Requires fine-tuning, may be prone to overfitting without regularization.

Neural Networks (Deep Learning): Use case: Complex tasks like image recognition, natural language processing, and sequential data analysis. Strengths: Can model highly complex and nonlinear relationships, state-of-the-art performance in many domains. Considerations: Requires large amounts of data, computational resources, and expertise in model tuning.

Support Vector Machine (SVM): Use case: Classification tasks, especially in situations with clear class separation. Strengths: Effective in high-dimensional spaces, works well with limited data. Considerations: May not perform well on large datasets, sensitive to parameter tuning.

Time Series Forecasting: Prophet: Developed by Facebook for forecasting with daily observations and seasonal effects. ARIMA: Suitable for univariate time series forecasting. LSTM and GRU: Deep learning models for sequential data with long-term dependencies.

Time Series Models: ARIMA (AutoRegressive Integrated Moving Average): Suitable for time series forecasting of variables like temperature and air quality. Prophet: Developed by Facebook for forecasting with daily observations and seasonal effects. LSTM and GRU: Deep learning models for sequential data with long-term dependencies.@

Anomaly Detection: Isolation Forest: Detects outliers and anomalies in data. One-Class SVM: Useful for one-class classification, where you want to identify data points that don't conform to the norm.

Ensemble models: Combine different models