benjaminweymouth / housing-market-forecasting-time-series-analysis

This repo uses Natural Langauage Processing, time series analysis, and ARIMA to explore predictive housing trend analysis.

Canadian Housing Market Forecasting: Time Series Analysis

Introduction

This jupyter notebook will utilize VAR, VARMAX and auto-arima to look at the long-term trends in the Canadian housing market to give us some indication of the future trends. VAR, VARMAX and auto-ARIMA will provide predictive modelling using multivariate analysis, similar to the VAR analysis. However, this will be a different analysis using VARMAX.

Hypothesis

Our hypothesis for VARMAS will be similar to the VAR analysis. We will take quantitative data in order to make a conclusion for the future. We will utilize VAR and VARMAX: two different models to acheive an understanding of multivariate analysis and how to pick the best predictive tool. Our hypothesis is that the Canadian housing market will experience a downturn similar to the American Housing Crash of 2008.

DataSets and Feature Selection

In order to prove our analysis we will utilize various sources. We will utilize data from the Bank of Canada, from Statistics Canada and from news sources. This Jupyter Notebook specifically will focus on the second of two multivariate forecasting models: VARMAX with Auto ARIMA.

For our feature selection we will use Residential Mortgage figures, The Housing Affordability Index, Prime rate figures and figures outlining the 5-year Conventional mortgage in Canada.

Conclusions: Quantitative, Qualitative and Overall Trends

Quantitative Conclusion:

The quantitative conclusion is that the current large trends will continue with some peaks and valleys. In short, since the Housing Affordability Index predicted to go up- this means that it will be more difficult for Canadians to afford a home. Both VAR and VARMAX models support this conclusion

Qualitative Conclusion:

The Negative sentiment articles in the canadian press ros significantly during the last period corroborating our quantitative results of a an imminent change in the actual trends, there is a significant worry about the change in the Canadian Housing Market especially with the retraction of the low interest rates introduced during Covid by Bank of Canada making mortgages higher and driving the loans rates to a newer high.

Overall Conclusion:

We conclude that from both pipelines, there is a an imminent change coming which would align with the downturn that was in our hypothesis. We believe that the three pipelines in our analysis shown in our Jupyter notebooks support this conclusion.

Jupyter Notebooks (3 pipeline analysis)

NLP Analysis: Click for NLP

VAR Analysis: Click for VAR

VARMAX with Auto-ARIMA: Click for VARMAX

PowerPoint Slides (Analysis and References)

Download the PowerPoint: Click for PPT