Forecasting of S&P House Price Index using Autoregressive integrated moving average (ARIMA) methodology.
Check/Run the scripts in the following order:
-
DataImportFromWeb.R - this imports Housing Price Index data from the webpage
-
DataMunging.R - performs basic data cleaning, reshaping and validation
-
Basic_Overall_Graphs.R - Generates timeseries plot for all the cities.
-
AutoARIMA.R - a. In this script, Stationarity of timeseries have been check using Stationarity test such as shapiro Test; Dickey-Fuller Test; Kwiatkowski-Phillips-Schmidt-Shin (KPSS) test; ACF- PACF Test. This also tells the order of differencing required for each TS. b. autoarima function is used, predicitons were generated too.
-
ARIMA_Model_Testing_and_Fitting.R - Here, ARIMA model is tested on different values of P-D-Q for each timeseries. Models were fitted on the best performing orders of P-D-Q.
-
ARIMA_Model_Diagnostic_Checks.R - Diagnostic checks (Ljung-Box Test, qqNorm Test) are performed on the fitted models.
-
TimeSeries_Decomposition_Trend_Seasonality.R - Timeseries is decomposed into Seasonality, Trend and Remainder and plotted.
-
ARIMA_Model_Forecasting.R - Predictions for the next 18 months are created and plotted.
-
ARIMA_Prediction_Plots.R - Final prediction plots all together
-
Model Architecture of the project at : https://github.com/jssasan/ARIMA-Time-Serie-Analysis-of-SandP-Housing-Price-Index/blob/master/ARIMA_Modeling_Architechture.png
-
Quick Insights of the project at : https://github.com/jssasan/ARIMA-Time-Serie-Analysis/blob/master/ARIMA_S%26P.pdf