‘Cab sharing’ is a well known concept of public transport service that can enable people to use taxi services at low cost. Primarily this service is targeted to achieve two broad goals. One is availability of an economically viable cab service option to areas which are poorly serviced by public transport wherein the only other alternative would be a high cost cab service. Shared cabs could cater to those commuters who travel frequently on highly congested routes and often face difficulties because of unavailability of good transport facility. Secondly cab sharing facility can help the cab service provider maximize revenue by catering any untapped customers, reducing per trip cost and avoiding any customers due to unavailability of cabs.
With roaring fuel prices and the need to maintain a high fleet of cabs several financial and operational challenges are faced by cab service providers including our client “Your Cabs”. The biggest challenge a cab service provider faces is to maintain the right tradeoff between capturing maximum consumer surplus from the customers to whom cab service can be provided and between losing out on revenues from customers either due to unavailability of cabs or due to offered rate cards which are higher than customer’s willingness to pay.
Also another operational issue which cab service providers face is variability of demand. Different routes at different time slots of the day have varied demand for the cabs. A model which can predict approximate number of cabs required from a particular route at a particular time would help cab service providers distribute or allocate resources in a more informed way.
Exhaustive data set that was made available provided us with data pertaining to details like the number of cabs booked from a particular area to a particular destination. Analysis of the data that was made available helped us gather insights such as routes such as (Airport, Whitefield, Marathahalli) that witness high traffic. Further analysis of sub set of the above data provides details showing a definite pattern in how bookings are done on a particular day or a particular time slot within a day.
Visualization of the data followed by application of data analytics such as classification techniques ‘KNN’ and ‘Naïve-Bayes’ would help us predict the demand levels among different routes. Further this prediction if compared to a pre decided demand limit can help cab service providers locate all congested routes that have a possibility of optimizing per trip cost by offering cab sharing.
With the designed data model, help desk personnel will have necessary data points to offer customers requiring a cab service the option to avail shared cab offered at lesser price (if the data model predicts high demand on the customer specified route/day and time). Also the prediction model can help cab service reduce operational costs by exploiting the possibility of offering shared cabs service and by maintaining an optimum cab fleet for a particular route thereby reducing the overall maintenance cost.