Machine Learning Fundamentals
The ML concepts data science interviews probe — bias and variance, models, metrics, features and validation.
This assessment checks the reasoning a data scientist or ML engineer needs before writing any code: framing problems as supervised or unsupervised, the bias–variance trade-off and regularisation, how common algorithms behave, choosing the right metric for imbalanced data, sound validation without leakage, feature engineering, and the basics of clustering, dimensionality reduction and neural networks.
What it measures
Core Concepts
Problem framing, overfitting, bias–variance and regularisation.
Supervised Learning
Regression, classification, trees, ensembles and k-NN.
Model Evaluation
Precision, recall, F1, ROC-AUC, RMSE and cross-validation.
Data & Features
Splits, leakage, scaling, encoding and missing data.
Unsupervised & Deep Learning
Clustering, PCA, neural network training basics.
Who it's for
- Aspiring data scientists and ML engineers
- Analysts moving into predictive modelling
- Students preparing for data science interviews
How the test works
- 20 questions drawn at random from a larger bank, balanced across every skill area above.
- Answer options are shuffled for each candidate, and the timer keeps running once you start.
- Your answers save automatically. If your connection drops, reopen your link to continue.
- Score 70% or more to earn a verifiable ARISE certificate. You get 2 attempts.
Try a sample question
A model scores 99% on training data but 70% on new data. What is the most likely problem?