TS models project report.docx - Gold Price Regression...

Gold Price Regression Models
Apply egression models using data partitions. Consider the
following 5 regression-based models:
i. Regression model with linear trend
ii. Regression model with quadratic trend
iii. Regression model with seasonality
iv. Regression model with linear trend and seasonality
v. Regression model with quadratic trend and seasonality.
vi. Holt Winter
a. Regression Model with Linear Trend:
-> The regression model with linear trend follows the following equation:
y
t
= 25781.4 + 103.5 t
-> Regression with Linear Trend model using summary () function is shown below:

forecast() function:
This data is a good fit for the train data(train.ts) since it has a multiple R-squared value of 0.5319 and
adjusted R-squared value 0. 5252.We can infer from the summary that this model is statistically significant.
Thus, we can use this model for time series forecasting.

b. Regression Model with Quadratic Trend:
-> The regression model with quadratic trend follows the following equation:
y
t
= 73800.248 + 2009.845 t – 20.823 t^
2
-> summary():
->forecast():
->This data is a good fit for the train data(train.ts) since it has a multiple R-squared value of 0.8692 and
adjusted R-squared value of 0.8638.
->The value of F-statistic is statistically significant since p-value is lower than 0.05.