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

79 practiced

Macroeconomic covariates like fuel price or CPI are often monthly while sales are daily. Describe strategies to align these exogenous series with daily forecasts, how to include exogenous forecast uncertainty, and pitfalls such as lookahead bias when using future macro forecasts as features.

MediumTechnical

97 practiced

Explain strategies to produce and validate prediction intervals for LSTM-based forecasts. Cover approaches such as quantile loss, Monte Carlo dropout, deep ensembles, and bootstrap residual sampling. Describe validation steps you would take to choose an approach for production including calibration checks.

EasyTechnical

96 practiced

What is Prophet (by Facebook/Meta) for time-series forecasting? Describe its model components including trend changepoints, yearly/weekly/holiday seasonality, how it handles missing data and outliers, which types of business time series it performs well on, and limitations a BI analyst should be aware of when using it for operational forecasts.

HardTechnical

90 practiced

Describe methods to handle temporal aggregation mismatch where sales are recorded weekly but business requires daily forecasts. Discuss disaggregation techniques such as proportional allocation using intra-week patterns or auxiliary covariates, interpolation approaches, how to propagate uncertainty from aggregated to disaggregated forecasts, and how to evaluate disaggregated predictions without daily ground truth.

MediumTechnical

69 practiced

Given a table daily_sales(store_id, sku_id, sale_date, sales_quantity), write Python (pandas) code to create lag features (lags 1, 7, 14), rolling means (7 and 30 days), and cyclic encodings for day-of-week and month. Explain how to avoid data leakage when generating features for training and how to prepare features for forecasting multi-step horizons.

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.