Kanya.Life uses data science to provide sharper insights on female infanticide in India across cities, towns, and villages. This project also tries to break the myth that, somehow, this problem is limited to remote, economically backward, rural areas. The initiative uses large and openly accessible datasets provided by the Government of India's Office of the Registrar General and Census Commissioner, the Python programming language for adjusting the structure of the data as well as for visualizing it, and Plotly for the creation of interactive graphs. See the problem, data, and insights pages for more information.
Bibliographical Source Citations
Asian Centre for Human Rights. Female Infanticide Worldwide: The Case for Action by the UN Human Rights Council.
Transcend Media Service. Asian Centre for Human Rights, June 2016. Web. 5 June 2017.
DTE Staff. "India Witnesses One of the Highest Female Infanticide Incidents in the World: Study." www.downtoearth.org. Down to
Earth, 8 July 2016. Web. 5 June 2017.
"Female Infanticide." BBC. BBC, 2014. Web. 5 June 2017.
"Sex Ratio." World Health Organization. World Health Organization, n.d. Web. 5 June 2017.
"The Worldwide War on Baby Girls." The Economist. The Economist Newspaper, 06 Mar. 2010. Web. 5 June 2017.
All images were downloaded from Unsplash and Pixabay, websites with a generous photography community that provides pictures free of copyright and other restrictions.
About Me
My name is Tarun Amarnath, and I am a high school student at Saint Francis High School in Mountain View, CA and a resident of Cupertino. I created Kanya.Life because I have a deep interest in applying data science to address large problems in the society. Four of my previous research projects are below:
I am a co-president of my school's robotics club and an active member of the programming club, having attended multiple hackathons and programming competitions. In addition, I compete at speech & debate competitions every month. In my spare time, I enjoy reading, art, and playing basketball with my brother.
- San Francisco Crime Prediction, 2016: San Francisco bustles with technology and innovation yet also suffers from widespread criminal activity. I worked on an award-winning project that uses predictive analytics and machine learning to predict, without bias, the category of crime that is likeliest to occur at a certain time and place in San Francisco. The data contained information about two million past crimes that have taken place over the past 13 years. I was invited by a venture capital firm in Palo Alto and a Microsoft research team to present my project, which was also selected for the California Junior Science & Humanities Symposium.
- Diagnosing Schizophrenia, 2017: I worked on a data science project to diagnose schizophrenia using multimodal brain imaging data. Numeric data collected from MRI brain scans can be analyzed to determine whether a patient has schizophrenia. The data includes Functional Network Connectivity values, correlation values between brain maps of the same person over time, and Source-Based Morphometry loadings, which describe patterns of brain structure related to gray matter values. The model generated an accuracy of about 80%, a value far better than the baseline of random guessing. This has also been published in Microsoft’s Cortana Intelligence Gallery and has been viewed or downloaded by over 900 data scientists around the world.
- With advances in machine learning, computers can take in doctor’s notes and interpret them to return a diagnosis of a problem using feature extraction methods such as n-gram and algorithms such as SVMs or decision trees. Working at Stanford University's Biomedical Informatics Research Center, I, along with a partner, looked into these methods in classifying patients into four groups of smokers and determining the presence of obesity through textual analysis of Electronic Health Records. Ultimately, we implemented a new combination of feature extraction and classification, using modified frameworks of Doc2Vec and Gradient Boosted Decision Trees, to create a supervised learning classifier returning a precision of 91%. This method can then be applied to further classification problems, such as problems arising from infections or classification of certain intensities of cancer.
- During a paid summer internship at the Beth Israel Deaconess Medical Center (the teaching hospital for Harvard Medical School), I worked on creating a pipeline that would help ease the process of researchers in identifying the impact of anti-miRNA molecules on pancreatic cancer cells. This would significantly reduce the amount of time required to analyze the samples taken from mice used by Dr. Frank Slack and his team, whose lab I worked at.
I am a co-president of my school's robotics club and an active member of the programming club, having attended multiple hackathons and programming competitions. In addition, I compete at speech & debate competitions every month. In my spare time, I enjoy reading, art, and playing basketball with my brother.