This project provides an insight into Time series forecast on hourly energy consumption data from 2002 to 2018 using XGboost and Prophet
The aim of this project is to carry out time series forcast on energy data using two models and then compare the accuracy of both models.
The data is an hourly distribution of energy consumption for 17 years in a city and this was made available by Rob Mulla on Kaggle with full permission for whoever wishes to use it.
- The accuracy metrics for XGBoost proves to be way better than that of Prophet.
- Prophet is more explicit when it pertains to plotting components and seasonality in time series data The above conclusion is only valid for the above data using the same methodology and is in no way a generalization of the accuracy of Prophet model as compared to XGBoost.
Carry out cross validation on the data using Prophet to observe the effect on all the accuracy metrics.