Projects
Short-term Demand and Supply Forecasting for Bike Taxis
Bike taxis have become a popular transport option, nowhere more so than in Indonesia where consolidating bike taxis into an app-based platform resulted in the first tech unicorn in the country. However, predicting the fluctuating demand and corresponding in a large city is challenging - especially since the desire for bike rides is strongly connected to the weather in tropical regions. A model to forecast the local demand and supply for the upcoming fifteen minutes was requested, based on current and historic data. The problem was divided into segments: the demand was predicted based on historical information, current trend, and weather information; the supply was predicted based on the historical driver tendencies, the current locations of drivers, and the estimated endpoints of currently active rides. Neural networks and Markov chains were employed to arrive at a model with around 90% forecast accuracy.
Modeling Loan Default Risk from Subjective Expert Experience
Our client, a microfinance company that disburses home loans to low-income individuals working in the unorganized sector, required a system to rate the risk of such loans. The risk of default is high because the applicants’ creditworthiness is not documentarily established, and the loan duration is for a considerable period of time. Sales personnel gather a multitude of information, either directly from the applicant or from different sources. This information becomes the primary data source to analyze and establish whether the loan will be approved, based on an estimation of the probability of successful repayment by the applicant. The principles of Aashiyan have decades of experience in underwriting microfinance loans and are able to make considered decisions based on the available data. However such expertise is mostly informal and subjective and cannot be easily translated into policy or automated risk-scoring. Each evaluation is time-consuming and thus makes it expensive for the organization to scale. Additionally, in the absence of a formal mathematical model, new information is harder to incorporate, and the evaluations suffer from a lack of flexibility. Our system formalizes the subjective expertise of the principals and builds a mathematical structure for risk scoring. The model then was integrated into the client’s data collection system and was able to instantly provide risk-scores as the information was entered. Finally, the model is made adaptable to information about loan outcomes and is able to learn from experience.
Products
Optimal Pickup and Dropoff of Employees by a Fleet
A large call center in a major Indian city needed to optimize the use of its fleet of vehicles to pick up and drop off its employees prior to and on completion of their shifts. Since the workforce usually had an absence rate of 25%, a static solution to the problem was not feasible. Along with cutting costs, employee satisfaction was a major concern. We mapped the situation to a mathematical optimization problem where the roster of the shift was to be partitioned into vehicles and the route of the vehicle depended on the order of employee dropoffs. The additional time an employee spends in the vehicle, beyond the time needed for a direct dropoff, was considered a cost in terms of employee satisfaction. The optimization problem was a combination of two NP-complete problems and as a result, was not amenable to a direct solution. Effective approximation was achieved by use of a genetic algorithm and the results decreased the cost to one-third while retaining employee satisfaction at current levels.
Consulting
Intra-City Freight Flow in the City of Kolkata
Fright transport around the city of Kolkata is complicated by the presence of River Ganges which requires traversal of the river across two main bridges resulting in congestion and delays. The World Bank Group intended to build roll-on/roll-off jetties at several locations to augment the available routes for trucks to cross the river. A study was commissioned to identify the impact of intra-city freight movement in the city of Kolkata as a result of these new crossings. The city was geographically divided into natural regions demarcated by canals or railway lines such that there were limited ways to cross from one region to the other. At each crossing, surveyors counted all goods transport vehicles while also recording their size and the type of goods carried. The survey lasted one week at each location, twenty-four hours a day. An additional survey was conducted at each large market in the city to interview shop owners to gauge and understand delivery patterns. The results were combined into a topological model which resulted in a visualization of the intra-city traffic flow by type of goods and vehicles across time and day of the week. As a result, the model could deliver a view of the potential consequences of local disruptions and deviations and was used to plan the location of the proposed jetties.
Route Rationalization for West Bengal Transport Department
The public transport network in Kolkata is vast and fragmented, being served by multiple government agencies and many private fleet owners. There was a need to consolidate routes and capacities among these operators to make the network efficient. To achieve this goal, each bus route was mapped by surveyors who marked the de facto trails and stops along with the number of passengers embarking and disembarking at each stop. This data was used to create a geographic and topological model of the bus network. Route sections were characterized and origin-destination level demand was inferred using Monte-Carlo techniques. Routes we then rationalized based on recommendations to merge, extend, or split routes using graph-theoretic measures. The results were presented to Transport Department officials and have since been implemented. This study was undertaken on behalf of the World Bank Group.
Research
Constructing Statistical Characteristics of COVID-19 Infection Trees from Contact-Tracing Data
The global impact of the COVID-19 pandemic has highlighted the need for modern data-driven approaches to managing such epidemiological events with minimum socioeconomic impact. To this end, we analyze publicly available contact tracing data from Karnataka, India to rebuild infection trees based on the ancestry of infections revealed in the data. In an attempt to statistically analyze these trees, we find that both the number of infections originating from a person as well as the size of the tree created by a hierarchy of such infections show a remarkably similar characteristic: the tails of these distributions decay slowly and appear to conform to the power-law form. As a consequence of this discovery, mitigation strategies could be designed by identifying and containing super-spreaders along with milder general restrictions.
Heterogeneous Contact Networks in COVID-19 Spreading: The Role of Social Deprivation
We have two main aims. First, we use theories of disease spreading on networks to look at the COVID-19 epidemic on the basis of individual contacts -- these give rise to predictions that are often rather different from the homogeneous mixing approaches usually used. Our second aim is to look at the role of social deprivation, again using networks as our basis, in the spread of this epidemic. We choose the city of Kolkata as a case study, but assert that the insights so obtained are applicable to a wide variety of urban environments that are densely populated and where social inequalities are rampant. Our predictions of hotspots are found to be in good agreement with those currently being identified empirically as containment zones and provide a useful guide for identifying potential areas of concern.