Feestdagen-/promotie-effectvariabelen maken

In de vorige oefening zagen we een aanwijzing voor seizoensinvloed, maar laten we testen of die er echt is. Je bazen denken dat hun producten meer in trek zijn rond de weken van Kerstmis, Nieuwjaar en Valentijnsdag. De marketingafdeling geeft ook aan dat ze de afgelopen 5 jaar in de week vóór Moederdag promoties hebben gedraaid. Laten we een binaire indicatorvariabele voor Nieuwjaar maken!

Deze oefening maakt deel uit van de cursus

Vraag naar producten voorspellen in R

Cursus bekijken

Oefeninstructies

Maak een datumindex voor de weken rond Nieuwjaar in je trainingsgegevensset en noem die n.dates.
Bouw een xts-object met de naam newyear met de datumindex n.dates.
Maak een reeks van 154 datums vanaf 19 jan. 2014 (het tijdsbestek van je trainingsgegevens) en noem deze dates_train.
Voeg het newyear-object samen met de trainingsdatums uit het vorige punt en vul alle ontbrekende waarden met 0'en.

Praktische interactieve oefening

Probeer deze oefening eens door deze voorbeeldcode in te vullen.

# Create date indices for New Year's week
n.dates <- as.Date(c("2014-12-28", "2015-12-27", "2016-12-25"))

# Create xts objects for New Year's
newyear <- as.xts(rep(1, 3), order.by = ___)

# Create sequence of 154 weeks for merging
dates_train <- seq(as.Date("2014-01-19"), length = ___, by = ___)

# Merge training dates into New Year's object
newyear <- ___(newyear, ___, fill = ___)

Code bewerken en uitvoeren

Deze oefening maakt deel uit van de cursus

Vraag naar producten voorspellen in R

SkillTag.level.intermediateSkillTag.label

4.8+

Begin de cursus gratis

When it comes to forecasting, time series modeling is a great place to start! You need to forecast out the future values of sales demand and a good baseline approach would be ARIMA models. In this chapter you'll learn how to quickly implement ARIMA models and get good initial forecasts for future product demand.

Exercise 1: Loading data into xts object Exercise 2: Importing data Exercise 3: Plotting / visualizing data Exercise 4: ARIMA Time Series 101 Exercise 5: auto.arima() function Exercise 6: Interpret auto.arima Exercise 7: Forecasting and Evaluating Exercise 8: Forecast function Exercise 9: Calculating MAPE and MAE Exercise 10: Visualizing Forecast Exercise 11: Confidence Intervals for Forecast

Economic theory has a lot to say about predicting values of demand. Obviously, external factors like price, seasonality, and timing of promotions will drive some aspects of product demand. In this chapter you'll learn about the basics around price elasticity models and how to incorporate seasonality and promotion timing factors into our product demand forecasts.

Exercise 1: Prijselasticiteit Exercise 2: Prijselasticiteit berekenen Exercise 3: Resultaten van elasticiteit interpreteren Exercise 4: Seizoens-, feestdag- en promotie-effecten Exercise 5: Feestdagen-/promotie-effecten visualiseren Exercise 6: Feestdagen-/promotie-effectvariabelen maken

Huidige oefening

Exercise 7: Regressie voor seizoens-/feestdagen- en promotie-effecten Exercise 8: Significante feestdag- of promotie-effecten?Exercise 9: Voorspellen met regressie Exercise 10: Maak toekomstige voorspeller-variabelen Exercise 11: Voorspel toekomstige vraagwaarden Exercise 12: Voorspellingen uit regressie visualiseren

Time series models and pricing regressions don't have to be thought of as separate approaches to product demand forecasting. They can be combined! In this chapter you'll learn about two ways of "combining" the information gained in both modeling approaches - transfer functions and forecast ensembling.

Exercise 1: Residuals from regression model Exercise 2: Calculating residuals from regression Exercise 3: Visualizing residuals Exercise 4: Forecasting residuals Exercise 5: auto.arima function Exercise 6: Forecasting residuals with time series Exercise 7: Visualizing residual forecasts Exercise 8: Transfer functions & ensembling Exercise 9: Combining residuals from regression & time series Exercise 10: Visualizing transfer function forecast Exercise 11: Calculating transfer function MAPE and MAE Exercise 12: ARIMA Forecasting Exercise 13: Ensembling of Forecasts

Everything up until this point deals with making individual models for forecasting product demand. However, we haven't taken advantage of the fact that all of these products form a product hierarchy of sales. Products make up regions and regions make up states. How can we ensure that our forecasts reconcile correctly up and down the hierarchy? In this chapter you'll learn about hierarchical forecasting and how to use it to your advantage in forecasting product demand.

Exercise 1: Bottom-Up Hierarchical Forecasting Exercise 2: Build time series forecast for new product Exercise 3: Build regression forecast for new product Exercise 4: Ensemble forecast for new product Exercise 5: Calculate bottom-up forecast for whole region Exercise 6: Top-Down Hierarchical Forecasting Exercise 7: Build time series forecast at regional level Exercise 8: Using average historical proportions Exercise 9: Using proportion of historical averages Exercise 10: Middle-Out Hierarchical Forecasting Exercise 11: Build time series forecast for new region Exercise 12: Top-down forecast for new region Exercise 13: Bottom-up forecast for whole state Exercise 14: Congratulations!