Time-Series-Modelling

TS_Exponential_Smoothing_Models

This project aims at building a time series model to create an hourly forecast of temperature for a retail store. The Store’s analysts believe that extreme outdoor temperatures may affect the sales of the main retail store in Harrisburg,PA; They want a forecast of these temperatures to help them further evaluate this claim.

The R script includes code for the following visualizations and models-

1. Time series decomposition
2. Plotting of the temperature series as a time series object
3. Fitting a trend line on the time series plot
4. Calculation of a seasonally adjusted data
5. Plotting seasonally adjusted values on the time series plot
6. Building a Single Exponential Smoothing Model
7. Building a Linear Expotential Smoothing Model
8. Building a Damped Trend Model
9. Building a Holt-Winters ESM
10. Building a Holt-Winters ESM- Multiplicative
11. Checking Accuracy Statistics on the test dataset

TS_Arima_Modelling

A retail store wants to build a weekly sales forecast model for two of its stores in - Phoenix and Tucson, AZ

The R script includes code for the following visualizations and models-

1. Creating time series object and plotting it
2. Building Linear Expotential Smoothing Model for both Tucson and Pheonix
3. Calculating Mean Absolute Percentage Error
4. Checking Stationarity through plots
5. Building Autoregressive Models for Tucson and Pheonix
6. ACF and PACF plots of residuals
7. White noise Tests and plots
8. Forecasting sales
9. Checking Accuracy on the validation dataset