Little Bits of Artificial Intelligence

Start
|
|—> 1. What is the Problem Type?
| |– Classification ——> Go to Step 2
| |– Regression ———-> Go to Step 2
| |– Clustering ———-> Use K-Means, DBSCAN, Hierarchical
| |– Time Series ———> Use ARIMA, LSTM, RNN
| |– Recommendation ——> Use Collaborative Filtering, Matrix Factorization
|
v
2. How Much Data Do You Have?
|– Small Dataset ————-> Prefer: Logistic/Linear Regression, Naive Bayes, SVM
|– Large Dataset ————-> Use: Random Forest, XGBoost, Neural Networks
|
v
3. Do You Need Interpretability?
|– Yes ———————–> Use: Logistic Regression, Decision Trees
|– No ————————> Use: Random Forest, Gradient Boosting, Deep Learning
|
v
4. What is the Feature Space Like?
|– High-Dimensional ———-> Use: Lasso/Ridge, PCA before model
|– Non-Linear Relationships –> Use: Tree-based, Kernel SVM, Neural Networks
|
v
5. Are There Resource Constraints?
|– Limited CPU/GPU ———–> Use: Lightweight models
|– Ample Resources ———–> Use: Deep Learning, Ensembles
|
v
6. Does the Model Overfit?
|– Yes ———————–> Add Regularization, Try Simpler Models
|– No ————————> Good, Continue
|
v
7. Real-Time or Offline Use?
|– Real-Time —————–> Use: Logistic Regression, SVM, Pruned Trees
|– Offline Batch ————-> Use: XGBoost, Deep Neural Networks
|
v
End → Final Algorithm Shortlist