Note: This post originally appeared on SocialStory January 15 2015
While most organisations know the problems plaguing them and affecting their progress, getting an outsider’s perspective always helps. In some cases out of the box thinking, based solely on data available at hand, does provide unique solutions to issues. With the aim of bringing the data science community together with mission driven organisations, DataKind Bangalore organized an event titled Project Accelerator Night on January 10, 2015. The goal was to assist non-profit organizations in taking their first steps towards a successful data science project. Each organisation briefly spoke about its functioning, the problems it was solving and the issues it faced. Then the participants broke up into different focus groups and brainstormed how Datakind could get leading attrition indicators using data science and offer solutions to these problems Here are the organisations and the discussions that ensued:
Teach for India(TFI) started out in 2009 as a nationwide movement of college graduates and young professionals committing two years to teach full-time at under resourced schools and working towards the pursuit of equity in education. Currently, Teach for India is active in seven cities, with a network of 910 fellows and 660 alumni.
Sahil Manekia, who looks after organizational effectiveness, represented the organisation. He explained how TFI gathered data from monitoring their dashboards, collecting data on fellow attritions, clustering Fellow Engagement Surveys and how they measured their impact by comparing data from schools where TFI is active and where it is not. TFI’s flagship fellow program is supported by a full time staff of nearly 170. Mumbai, Delhi and Pune are the primary staff locations.
Problem Statement: Attrition among the staff is a growing concern for TFI. While the staff is compensated competitively compared to other educational non-profit peers, it’s not competitive when compared to the for profit sector. Other personal reasons could be leading factors for staff attrition. Even though the number of people leaving might seem small, it is often tough to get qualified and committed staff members on board for a long period.
Proposed Solutions: The discussion was led by meet up ambassadors Suchana Seth and Udaya Chitta, Some of the data available came from schools and subjects taught, dropout data in classes, teachers’ performance reviews and exit interviews. The idea was to see how TFI was making a difference, as seen from the student rolls, and to see if successful models could be built and replicated across a larger presence. The other area of discussion was measuring dropout rates in schools and measuring TFI’s impact in the schools where its fellows are teaching. They concluded that the bulk of the data could be collected from the census to segment the schools by demography. They further discussed how TFI could incorporate spatial segmentation with existing metrics.
Website: Teach For India
Digital Green, founded in 2008, is an international, non-profit development organization that builds and deploys informative videos to amplify the effectiveness of development efforts to bring about sustained, social change. Their model combines technology and social organization to maximise the potential of building the capacity of community members on improved, sustainable agriculture, livelihood and health interventions.
Aadish Gupta, who takes care of predictive analytics at Digital Green, represented the organization at the event. He explained processes involved in conceptualizing the content of the videos, pre and post production and how it was done with the help of mediators and partner organisations. The videos, with average run times of 8-10 minutes, were shot using locally available resources and broadcast at village gatherings through pico projectors.
Problem Statement: Digital Green has a repository of 4000+ videos based on agriculture, health and hygiene; and nutrition. Aadhish wanted to explore how they could go about building a recommendation engine, to recommend videos to be screened in about 9000 villages across India.
Proposed Solutions: The discussion was spearheaded by meet up ambassadors Althaf Kandi and Gaurav Godhwani, and the group brainstormed how they could map existing villages with open geospatial data and state-wise agriculture data so that they could build key identifiers for each village. Then they dived into understanding how to rank similar videos by quantifying video adoption rate from past video screening process data. Later, they discussed technical details of this recommendation engine, and how its feedback could be recorded to improve its performance.
Website: Digital Green
Janaagraha was established in 2001 as a non-profit organisation that works on combining the efforts of both the government and citizens to transform Indian cities and ensure a better quality of life. Their goal is to improve the quality of urban infrastructure and services; and quality of citizenship. Their civic portal IChangeMyCity promotes civic action at a local neighbourhood level and sister portal IPaidABribe encourages people to speak up against corruption and report instances where they meet honest officers or are asked for bribes and also includes a bribe helpline.
For the discussion though, the focus was only on IChangeMyCity. Sunil Nair and Rajith Shaji, who lead and take care of managing and ensuring effectiveness of the web portal, represented the organisation during the event. They stated that though there are 1 lakh users on their database, they receive only about 16k complaints a month, and manage to resolve half of those.
Problem Statement: They discussed how similar complaints in same geographies coming from different users could be combined into a single complaint and complaint priority could be re-assigned. They also discussed how they could devise a system to predict probability and time for an issue to get resolved along with a trends engine to get issues trending in a locality at a given time.
Proposed Solutions: The discussions were steered by meet up ambassadors Venkatramanan PR and Vinod Chandrashekar. They suggested how recommended landmarks could be used for clubbing complaints and identifying key metrics for analysing past data about complaints and their resolvement timelines. The group further focused on understanding key contributing factors for building the trends engine.
Website : IChangeMyCity, IPaidABribe
Karnataka Learning Partnership (KLP) is a public platform where a network of non-profits working in education bring their data together to analyse and present factual assessment regarding the public schooling system in Karnataka. All stakeholders involved in primary education can participate on this collaborative platform and contribute to the cause of ensuring better schools and education in Karnataka.
KLP representative Bibhas Ch Debnath walked the attendees through the data, analysis and reports available via various APIs. He mentioned that several non-profits already use their data in multiple ways to drive change in primary education effectiveness and education policy.
Problem Statement: As KLP themselves are a data provider, they did not have any immediate data science issue but discussions were centered on different ways organisations and individuals could engage with KLP’s resources.
Proposed Solutions: DataKind and KLP representatives discussed how they could work together and help each other in pooling in resources and networks. KLP promised to help DataKind reach out to various non-profits who use KLP data as they would most likely have data science requirements. Some of the problem statements from other NGOs require data regarding child education in rural areas. DataKind could explore the use of KLP data for those requirements.