Forecasting
54 bites tagged Forecasting — interview questions with model answers, and 60-second explainers.
Transform a time series for a supervised learning model?
This tests your ability to reframe a time series problem for tabular models. A great answer explains creating features from lags, rolling windows, and calendar data, then emphasizes using a time-aware validation split. A red flag is forgetting validation.
Describe two methods for generating prediction intervals
This tests your understanding of forecast uncertainty. Describe two methods: 1) assuming normally distributed errors and using a standard deviation multiplier, and 2) bootstrapping residuals to simulate future paths.
Forecasting inventory with trend and weekly seasonality?
This tests mapping a business problem to a statistical tool. A good answer names Holt-Winters, explains its level, trend, and seasonal components, and discusses additive vs. multiplicative seasonality.
Explain stationarity in a time series
This tests your grasp of core time series modeling assumptions. A strong answer defines stationarity (constant mean/variance), explains its importance for ARIMA (stable patterns), and names a test (ADF) and a fix (differencing).
How would you measure a sales forecast model's accuracy?
This tests your ability to connect statistical metrics to business impact. A great answer defines MAE (linear error cost) and RMSE (penalizes large errors), explains the choice depends on business context, and stresses using a test set.
How do you build a Monte Carlo project forecast?
Tests your ability to model uncertainty and communicate probabilistic outcomes. A good answer covers gathering cycle time data, running thousands of simulations, and presenting results as a probability distribution (e.g., "85% confidence by X date"), not a…
How do you provide a probabilistic forecast for completing 25 stories?
This tests your ability to move beyond single-date estimates to probabilistic forecasting. A good answer outlines using historical cycle times in a Monte Carlo simulation to generate a range of dates with confidence levels.
Distinguish Throughput from Velocity in agile planning
This tests your grasp of outcome (Throughput) vs. effort (Velocity) metrics. Define both: Throughput is item count/time, Velocity is points/sprint. Contrast them by explaining Throughput measures actual delivery, not estimates.
How would you create a probabilistic forecast for a backlog?
This tests your grasp of probabilistic forecasting over single-date estimates. A good answer explains using historical throughput to run a Monte Carlo simulation, then presenting a range of dates with confidence levels (e.g., 50%, 85%).
Right-Shifting Forecasts: Why Your Deadlines Keep Moving
A forecast is like a GPS ETA in traffic; new tasks are like accidents ahead, pushing your arrival time further out. This happens when initial estimates are treated as fixed deadlines, ignoring new scope. The footgun is anchoring on the first date given.
Throughput: Measuring What Gets Done
Throughput measures how many work items a team *finishes* in a time period, not how busy they are. It's used for forecasting future work and spotting bottlenecks. The footgun: never compare throughput between different teams, as item sizes and context vary.
Time-Series Forecasting: Predicting the Future from the Past
Time-series forecasting uses past data points, ordered by time, to predict future values. It's used for capacity planning and financial modeling. The footgun is assuming past trends will hold, as sudden system changes can invalidate all predictions.
Forecasting with Monte Carlo Simulation
A Monte Carlo simulation forecasts outcomes by running thousands of 'what if' scenarios with random inputs. Use it to model complex systems like user growth with variable conversion rates.
Prophet: Time Series Forecasting for Seasonal Data
Prophet treats a time series as a sum of its parts: a long-term trend, seasonal cycles, and holidays. It's used for business forecasting, like predicting sales, when you have strong seasonal data. The footgun is using it for non-seasonal data.
The Bass Diffusion Model: Innovators vs. Imitators
The Bass Diffusion Model splits product adoption into innovators who buy first and imitators who follow the crowd, creating the classic S-curve of growth. It's used to forecast sales for new products by modeling how word-of-mouth drives adoption.
Linear Regression: Forecasting with a Straight Line
Linear regression forecasts the future by drawing a straight line through past data. It's used to predict outcomes like sales based on ad spend or energy use based on temperature.
Training vs. Test Sets: Judging a Forecast's True Accuracy
Don't judge a forecast by how well it fits the data it was trained on. To measure true accuracy, split your data into a training set for model fitting and a test set for evaluation, using data the model has never seen before.
Sales Velocity: How Fast Your Pipeline Makes Money
Think of sales velocity as the speedometer for your revenue engine, measuring how quickly your pipeline generates money. Sales leaders use it to forecast revenue and find bottlenecks.
Ensemble Forecasting: Predicting with a Crowd of Models
Instead of one 'best guess,' ensemble methods generate many forecasts to map the range of possibilities. This is crucial for complex systems like weather prediction, where a single model is misleadingly precise.
ARIMA: Forecasting by Modeling Autocorrelation
ARIMA models forecast a time series by learning its "memory"—how past values influence the next. It's used for forecasting sales or server load where patterns are driven by internal dynamics.
Prophet: Automated Time Series Forecasting
Prophet treats forecasting like decomposing a signal, breaking a time series into trend, seasonality, and holiday effects. It excels at predicting business metrics with strong seasonal patterns, like web traffic.
Time Series Cross-Validation
Unlike standard cross-validation that shuffles data, time series cross-validation respects the arrow of time. It evaluates a forecasting model by training only on past data to predict a future point, then rolling forward. The footgun is using random k-folds.
ACF vs. PACF: A Time Series Signal and Its Echo
Autocorrelation (ACF) measures a time series's total correlation with its past, including indirect echoes. Partial Autocorrelation (PACF) isolates the direct correlation by removing intermediate effects. They help select forecasting model parameters.
Holt-Winters Exponential Smoothing
Forecast time-series data by giving more weight to recent events. Holt-Winters smoothing is used to analyze metrics and make predictions by accounting for user-defined assumptions like seasonality.
Get Forecasting bites daily.
Five a day, five minutes, offline. With quizzes so it sticks.
Open testing — you’ll join as an early tester.