Skip to content

Latest commit

 

History

History
200 lines (119 loc) · 3.2 KB

weekly_plan.md

File metadata and controls

200 lines (119 loc) · 3.2 KB

Week 1

Tutorial

  • Welcome and course info (10 min)
  • Assignment 1 (15 min)
  • tsibble: index, key (25 min)

Workshop

  • tsibble: index, key exercise
  • Creating a tsibble from a csv
  • dplyr manipulation: summarise, filter

Week 2

Tutorial

2.10: Q1 (2 series), Q2, Q3, Q7

  • tsibble: data from files, frequency, interval, temporal gaps (20 min)
  • autoplot() time plots (15 min)
  • time series patterns: trend and seasonality
  • Time series graphics: season, subseries (15 min)

Workshop

2.10: Q9, Q10, Q11

  • Exam question B.1 from 2024 and 2023
  • time series patterns: difference between seasonality and cyclicity
  • lag plots, ACF graphs
  • cross-sectional visualization

Week 3

Tutorial

3.7: Q2 (2 series), ~Q3 (show, not ask), Q4 + STL decomposition on project data

  • Log and box-cox transformations (20 min)
  • STL decomposition (30 min)

Workshop

3.7: Q9, Q10 (add Q3 in detail into workshop too)

  • Other transformations
  • Other decompositions
  • Decomposition plotting
  • Exam B.2 from 2024

Week 4

Tutorial

5.11: Q1, Q3, Q5

  • Forecasting workflow -> tutorial setup (5 min)
  • Benchmark forecasting: mean, naive, snaive, drift (25 min)

Workshop

5.11: Q2, Q11

  • Linear trends, dummy seasonality
  • Decomposition forecasting

Week 5

Tutorial

5.11 Q8

  • Residual diagnostics (20 mins)
  • Accuracy evaluation: train, test, cross-validation (30 min)

Workshop

Ex 5.11: Q6, 7, 9, 10,

  • Exam Q A3 from 2024, and similar.

Week 6

Tutorial

8.8: Q1, Q2, Q3, Q4

  • Exponential smoothing concepts (20 min)
  • ETS forecasting (30 min)

Workshop

8.8: Q16, Q17

  • ETS forecasting by hand (30 min)

Week 7

Tutorial

8.8: Q5, Q8 (added), Q14

  • Intuition of exponential smoothing (10 mins)
  • Relation to benchmark models (5 mins)
  • Additive/multiplicative error, trend, seasonality (15 mins)
  • ETS forecasting (20 mins)

Workshop

8.8: Q10

  • Recall residual diagnostics (10 mins)
  • Dampened trend (10 mins)

Week 8

Tutorial

9.11: Q2, Q3(a,c), Q5

  • Stationarity (20 mins)
  • Transformation + differencing (10 mins)
  • ACF + PACF (20 mins)

Workshop

9.11: Q1, Q4, Q3(b)

  • White noise -- moved to W6 residual diagnostics (15 mins)
  • Backshift notation (15 mins)

Week 9

Tutorial

9.11: Q7, Q9

  • Identifying AR/MA from PACF/ACF (20 mins)
  • Non-seasonal ARIMA modeling (20 mins)
  • Backshift notation with ARIMA (10 mins)

Workshop

9.11: Q6

  • Long term forecast behaviour c+d (10 mins)

Week 10

Tutorial

9.11: Q11, Q14 (+ writing out models)

  • Recap non-seasonal ARIMA (10 mins)
  • Seasonal lagged autocorrelations (10 mins)
  • Seasonal ARIMA models (20 mins)
  • Backshift notation with SARIMA (10 mins)

Workshop

9.11: Q12, Q16

  • ARIMA forecasting by hand (20 mins)
  • STL+ARIMA forecasting (10 mins)

Week 11

Tutorial

7.10: Q6, Q4 (+ fourier)

  • Regression recap (5 mins)
  • Piecewise trends (20 mins)
  • Fourier seasonality (25 mins)

Workshop

7.10: Q1, Q2, Q5

  • Exogenous regressors (10 mins)
  • Scenario forecasting (10 mins)

Week 12

Tutorial

10.7: Q2, Q5

  • ARIMA errors and residual diagnostics (20 mins)
  • Dynamic harmonic regression (15 mins)
  • Project assessment support (15 mins)

Workshop

10.7: Q6

  • Equation writing (10 mins)
  • Manual forecasting (20 mins)