Machine learning basics: practice, test and prove it

Covers learning types, train/test splits, overfitting, bias-variance, core algorithms and evaluation metrics.

FreeNo sign-up to testNo attempts yet

Start the machine learning basics test Read the first lesson

Entry-level machine learning screenings focus on judgement more than maths. Typical questions ask which type of problem a task is, why a model scores well in training but poorly on new data, or which metric suits an imbalanced dataset. Some include a confusion matrix and ask you to compute precision or recall.

This page lets you practise those questions, with every calculation shown in the explanation. The lessons cover how to split and evaluate data honestly, and how to choose and read metrics for classification and regression.

Benchmark: The pass mark on this site is 80%, and machine learning basics are commonly screened for junior data scientist, ML engineer and data analyst roles. Requirements vary by employer.

Lessons

Splitting data, overfitting and the bias-variance trade-offHow train, validation and test sets work, how to spot overfitting and underfitting, and how cross-validation helps.Lesson 1

Choosing and calculating evaluation metricsAccuracy, precision, recall, F1 and RMSE explained with worked numbers, and how to pick the right one for the task.Lesson 2

Core algorithms: regression, trees, k-nearest neighbours and k-meansWhat each common algorithm does in plain terms, which problem type it fits, and the questions screenings ask about them.Lesson 3

Preparing features: encoding, scaling, missing values and avoiding leakageHow to turn raw columns into model inputs, why preprocessing must be fitted on training data only, and how leakage inflates scores.Lesson 4

Interview prep

Machine learning basics interview questionsMachine learning basics questions interviewers commonly ask, with answers that show how you reason about models.6 questions

Certificate

Free with an accountA PDF with your name, best score and date. Records this site's tests only.

Common questions

Do I need advanced maths for these questions?

No. You need basic algebra and the ability to calculate ratios such as precision and recall. Deeper maths helps later but is not what entry screenings usually test.

Are the questions tied to one library?

No. The concepts apply in any library. Where code appears, it is short and the idea matters more than the exact function name.

Does this cover deep learning?

Only at the level of general concepts such as overfitting and evaluation. Neural network architecture is outside the scope of this basics test.

Machine learning basics test for interviewWhat the screening round looks like, with a 5-question sample.