Time-Series-Forecasting-SARIMAX

Using SARIMAX for Time Series Forecasting on Seasonal Data that is influenced by Exogenous variables

(Prepared for) Sangam 2019 - ML Hackathon by IITMAA

Data Provided: Traffic Data (refer train.csv for more)

Data description

Columns	Description
date_time	Date, time, and hour of the data that is collected in the local IST time
is_holiday	Categorical Indian national holidays combined with regional holidays
air_pollution_index	Air Quality Index (10-300)
humidity	Numeric humidity in Celcius
wind_speed	Numeric wind speed in miles per hour
wind_direction	Cardinal wind direction (0-360 degree)
visibility_in_miles	Visibility of distance in miles
dew_point	Numeric dew point in Celcius
temperature	Numeric average temperature in Kelvin
rain_p_h	Numeric amount in mm of rain that occurred in the hour
snow_p_h	Numeric amount in mm of snow that occurred in the hour
clouds_all	Numeric percentage of cloud cover
weather_type	Categorical short textual description of the current weather
weather_description	Categorical longer textual description of the current weather
traffic_volume	Numeric hourly traffic volume bound in a specific direction

The traffic_volume attribute has to be forecasted on the basis of the time series data provided, taking the exogenous variables into account

Approach used: SARIMAX (Seasonal Autoregressive Integrated Moving Average with eXogeneous variables)

Reason: The data provided is seasonal, and it is a time series data with multiple exogeneous variables influencing the result. Hence, the optimal statistical model that can be applied to this task is SARIMAX

Main Modules Used:

• statsmodel package in Python