Problem Description:- The business problem involves the forecasting of Ottawa city bike traffic to understand the requirement of increasing bike corridors in the city. The traffic accidents in Ottawa is on rise, with a 7-year peak reaching in 2017. As per the Ottawa police, traffic accidents have increased by 23%. Additionally, the registered two-wheeler vehicles went up by 15000 with 5000 additional drivers.
The analysis of data across 6 junctions in Ottawa helped to solve the below two problems-
- Identifying the high traffic areas for bikers and small light-weighted vehicles, and planning for construction of special biker lanes
- Hiring traffic management personnel based on peak times
We check for the traffic during different months of the year for the last 5 years (4 years training + 1 year of validation) to forecast for the next coming years. The forecasting horizon is 1st Nov’17 to 31st Oct’18.
Performance Metrics:- Based on running various models on the training and validation data set, we arrive at the conclusion that “Holt’s multiplicative” is the best-fitted model. The model is performing better on all the metrics like MAPE, SSE etc.
Short Term: - Months of June and July see spikes in traffic for some select days. It can be used for the efficient allocation of traffic personnel
Long Term: - Creating corridor extensions for the junctions of OBVW, OGLD, SOMO.