Groupe d’études et de recherche en analyse des décisions

G-2016-76

Forecasting local warming: Missing data generation and future temperature prediction

et

Global warming is a much discussed topic as it sparks debate for shaping government policy and how humans should behave in reaction to climate change. Global warming can be considered at a local perspective when we look at temperature trends at an isolated region. In this work we aim to predict a local warming trend for Canada's capital city Ottawa, Ontario up to the year 2040 using optimization and machine learning techniques. We use data from the National Oceanic and Atmospheric Administration (NOAA) which archives historical weather data from approximately 9000 weather stations from around the world. Some of the datasets date back to 1955, however it is incomplete for a number of weather stations including Ottawa. In this work we first expand on a statistical based approach proposed by Robert J. Vanderbei to model local warming based on a previous day correlation. Then we present a forward algorithm which samples from the Laplace distribution to fill in the missing data. Lastly, we make predictions up to the year 2040 using the Neural Network toolbox within Statistica.

, 19 pages

Ce cahier a été révisé en novembre 2016