Practice the idea you just learned.
Each challenge is routed from a specific HumachLearn concept. No generic practice fallback: every button here tests the same idea you came to practise.
Trace the program state
After `x = [1, 2]` and `y = x`, what happens when `y.append(3)`?
Practise →Functions & Reusable ReasoningTrace the program state
After `x = [1, 2]` and `y = x`, what happens when `y.append(3)`?
Practise →Choosing Python CollectionsTrace the program state
After `x = [1, 2]` and `y = x`, what happens when `y.append(3)`?
Practise →pandas & Table ThinkingChoose the safe table operation
You need one row per customer before joining to a customer-level target. What should you verify first?
Practise →SQL Query FlowChoose the SQL operation
You need revenue by region after filtering cancelled orders. Which order is logically safest?
Practise →Excel as an Analytical ModelChoose the spreadsheet tool
You need a reusable import-clean-refresh workflow from a CSV. Which tool fits best?
Practise →Missing Data DecisionsChoose an imputation strategy
Income is skewed with strong outliers and 8% values missing. Which simple baseline is usually safer than mean imputation?
Practise →Scaling & DistanceSpot the preprocessing rule
Which workflow avoids scaling leakage?
Practise →Categorical EncodingChoose the safe table operation
You need one row per customer before joining to a customer-level target. What should you verify first?
Practise →Exploratory Data AnalysisRead a distribution
A long right tail makes which centre more robust?
Practise →Visualisation & StorytellingChoose a chart
You want to show the relationship between two numeric variables. Which chart is the natural first choice?
Practise →Train, Validation & TestChoose a legal split
Each patient has several recordings. What validation design protects patient-level independence?
Practise →Cross-ValidationChoose a legal split
Each patient has several recordings. What validation design protects patient-level independence?
Practise →Data LeakageChoose a legal split
Each patient has several recordings. What validation design protects patient-level independence?
Practise →Linear RegressionRead the residual signal
Residual spread increases as predictions increase. What pattern is this?
Practise →Logistic RegressionChoose the decision policy
A screening system must miss as few positive cases as possible. Which metric deserves special attention?
Practise →Decision TreesControl tree complexity
Validation error rises while training error keeps falling as depth increases. What should you do first?
Practise →Random ForestReason about an ensemble
Why does a Random Forest randomise both rows and candidate features?
Practise →Gradient BoostingReason about boosting
What does a new boosting learner usually focus on?
Practise →K-Nearest NeighboursChoose k
What is the usual effect of increasing k in KNN?
Practise →Support Vector MachinesInterpret C
Increasing C in a soft-margin SVM generally makes violations…
Practise →K-Means ClusteringChoose k responsibly
Which approach is better than selecting k only by eye?
Practise →Principal Component AnalysisChoose the next analytical step
Before choosing a sophisticated model, what should be clear first?
Practise →Classification MetricsChoose the decision policy
A screening system must miss as few positive cases as possible. Which metric deserves special attention?
Practise →Regression Metrics & ResidualsRead the residual signal
Residual spread increases as predictions increase. What pattern is this?
Practise →Bias, Variance & Learning CurvesRead the residual signal
Residual spread increases as predictions increase. What pattern is this?
Practise →RegularisationSpot the preprocessing rule
Which workflow avoids scaling leakage?
Practise →Hyperparameter OptimisationProtect evaluation
Where should hyperparameter search occur?
Practise →Probability Calibration & ThresholdsChoose the decision policy
A screening system must miss as few positive cases as possible. Which metric deserves special attention?
Practise →Explainability & Model InterpretationChoose the next analytical step
Before choosing a sophisticated model, what should be clear first?
Practise →Predictive UncertaintyRead the residual signal
Residual spread increases as predictions increase. What pattern is this?
Practise →Time-Series ForecastingChoose the next analytical step
Before choosing a sophisticated model, what should be clear first?
Practise →Neural Networks & BackpropagationTune learning rate
If loss oscillates or explodes immediately, which parameter is a prime suspect?
Practise →Attention & TransformersInterpret attention
What is safest to say about a high attention weight?
Practise →Graph Neural NetworksReason about message passing
After two GNN layers, a node can usually incorporate information from…
Practise →Deployment, Drift & MonitoringChoose the next analytical step
Before choosing a sophisticated model, what should be clear first?
Practise →