Machine Learning and Forecasting Algorithms Questions

An in-depth coverage of machine learning methods used for forecasting and time-series prediction, including traditional time-series models (ARIMA, SARIMA, Holt-Winters), probabilistic forecasting techniques, and modern ML approaches (Prophet, LSTM/GRU, Transformer-based forecasters). Topics include feature engineering for seasonality and trend, handling non-stationarity and exogenous variables, model evaluation for time-series (rolling-origin cross-validation, backtesting, MAE/MAPE/RMSE), uncertainty quantification, and practical deployment considerations such as retraining, monitoring, and drift detection. Applies to forecasting problems in sales, demand planning, energy, finance, and other domains.

HardTechnical

98 practiced

Propose feature-engineering and regularization strategies for high-cardinality categorical features (for example product_id) when training a global neural forecasting model. Discuss categorical embeddings, target encoding with temporal-leakage prevention, hash embeddings, dimensionality selection for embeddings, dropout/weight decay, and online update strategies for new categories.

EasyTechnical

100 practiced

Define probabilistic forecasting and give two practical methods to produce prediction intervals from deterministic models (for example: bootstrap residuals and quantile regression). Explain advantages of probabilistic forecasts for decision-making and give an example business decision where providing intervals or quantiles is necessary.

HardTechnical

98 practiced

Provide PyTorch code snippets to train a sequence-to-sequence LSTM for multi-step forecasting using teacher forcing. Include: data loader batch creation for tensors shaped (batch, seq_len, features), model forward pass producing multi-horizon outputs, loss computation across horizon, implementation of teacher forcing during training, optimizer step, and handling of variable-length sequences via padding and masks. Pseudocode is acceptable but include key implementation details.

EasyTechnical

93 practiced

Describe MAE, RMSE, MAPE, sMAPE and WAPE. For each metric explain sensitivity to outliers, interpretability for business stakeholders, and scenarios where one metric is preferred (for example intermittent demand, budgeting, capacity planning). Explain pitfalls of percentage-based metrics like MAPE when actuals include zeros.

HardTechnical

148 practiced

Your probabilistic forecaster's nominal 90% prediction intervals only capture 70% of actual observations (undercoverage). Describe a systematic approach to diagnose the source of undercoverage (model assumptions, heteroskedasticity, residual misspecification, covariate shift) and methods to fix it, including distributional modeling, heteroskedastic models, quantile models, bootstrap/conformal methods, and recalibration strategies.

Unlock Full Question Bank

Get access to hundreds of Machine Learning and Forecasting Algorithms interview questions and detailed answers.

Join thousands of developers preparing for their dream job.