MLiP_M5

Useful links and information:

Accuracy challenge

The materials from Brightspace:

• Online book about timeseries forecasting https://otexts.com/fpp2/what-can-be-forecast.html
• Methods for Intermittent Demand Forecasting https://www.lancaster.ac.uk/pg/waller/pdfs/Intermittent_Demand_Forecasting.pdf
• Mathematical background of scoring rules https://www.stat.washington.edu/raftery/Research/PDF/Gneiting2007jasa.pdf

More papers:

• LSTM for time series paper https://arxiv.org/pdf/1901.04028.pdf
• More LSTM https://dl.acm.org/doi/pdf/10.1145/3209978.3210006
• RNN + clustering https://arxiv.org/pdf/1710.03222.pdf

Tutorials:

• Predict Future Sales competition in Kaggle different algs preformance https://github.com/waylongo/predict-future-sale/blob/master/Readme.md
• Quite nice tutorial to get started with time series https://machinelearningmastery.com/time-series-forecast-study-python-monthly-sales-french-champagne/
• Classical forecasting methods cheat sheet https://machinelearningmastery.com/time-series-forecasting-methods-in-python-cheat-sheet/

Kaggle kernels:

• Basic EDA and predictions in python https://www.kaggle.com/tarunpaparaju/m5-competition-eda-models
• Extensive EDA in R https://www.kaggle.com/headsortails/back-to-predict-the-future-interactive-m5-eda

Kaggle discussions:

• A lot of tips: https://www.kaggle.com/c/m5-forecasting-accuracy/discussion/144067
• Practical NNs notes: https://www.kaggle.com/c/m5-forecasting-accuracy/discussion/144023
• M4 winner: https://www.kaggle.com/c/m5-forecasting-accuracy/discussion/133551

Uncertaintly challenge

The materials from bright space:

• Pinball loss function https://arxiv.org/pdf/1710.01720.pdf

The information collected by our coach.

• Nice article to start with. https://towardsdatascience.com/estimating-uncertainty-in-machine-learning-models-part-1-99fde3b0cbc1

I see the discussion section of the competition could be really helpful to understand the problem. Like following threads seems interesting to start with:

• https://www.kaggle.com/c/m5-forecasting-uncertainty/discussion/133613. Please go through the whole thread including discussion
• https://www.kaggle.com/c/m5-forecasting-uncertainty/discussion/133692
• Suggested in one of the comments. https://www.sciencedirect.com/science/article/pii/S0169207019301153
• https://github.com/Mcompetitions/M4-methods.

Understanding the uncertainty in weather/ meteorological forecasting would be worthy enough for this problem

• http://urban-computing.com/pdf/kdd19-BinWang.pdf
• https://www.researchgate.net/publication/328264084_Predicting_Weather_Forecast_Uncertainty_with_Machine_Learning.
• May be of your interest at some point. https://www.kaggle.com/fernandol/cracking-the-walmart-sales-forecasting-challenge