Moneycontrol PRO
Upcoming Event : LeapToUnicorn - mentoring, networking and fundraising for startups. Register now

Sima Taparia is a match for a data scientist

There is statistics hidden behind matchmaking that comes alive in 'Indian Matchmaking' on Netflix.

September 18, 2022 / 06:30 PM IST
Sima Taparia in Season 2 of 'Indian Matchmaking'.

Sima Taparia in Season 2 of 'Indian Matchmaking'.

The first season of Indian Matchmaking on Netflix left the audiences divided. Some called the show regressive. Others found it to be a reflection of Indian society. Matchmaking, however, is a skilled occupation. Matchmaker Sima Taparia, famously “Sima from Mumbai”, may have become a subject of memes after the show, but hidden behind her aunty-gaze are the methods of a data-scientist.

They say marriages are made in heaven. But there's no way to test this theory. Humans have tried to solve for spouse-selection, society’s oldest problem, in many ways. Romans drew lotteries while Indians drew kundalis. Few even believed they could sniff a partner out from their body odour. Pulse rates were recorded to measure physical attraction to a prospective match. All these invisible forces eventually made way for the matchmakers – a very visible intervention – for supply to meet demand and for the market to clear for marriage. As with other professions, computers soon arrived on the scene to do the matchmaker’s job, better and faster. But in India, where 90 percent of marriages are arranged even today, Sima aunty’s human touch has high demand among families.

For them, Sima is a match algorithm in flesh and blood. She collects data, to reduce information asymmetry in the market. Then she works with probabilities to optimise for a match. This is surprisingly close to how economists and statisticians view the world.

First you collect data

Sima aunty collects primary data. When asked about her method, she replied, “my work is to personally visit... understand personalities, understand preferences, see the standard of living, see their business, nature of girl and the boy”.

She tours through her client’s house, turning observations into variables for modelling. She makes the clients list out the ‘criteria’ they want their partners to fulfil.

After the visit, she is left to work with both structured and unstructured data. Concrete variables like age, height, location, income, ethnicity, and even skin-colour are easy to quantify for her. It’s the demands of finding someone flexible, adjusting or good-natured which become hard to measure.

Much like a data analyst, she keeps her data clean and organised in a ledger with rows and columns, a spreadsheet of her own, a database of bio-datas. Indian Matchmaking’s episode titles reflect her long-drawn method. The first episode seeks the “slim, trim and educated”. But by episode eight, matchmaking is reduced to “adjustment and compromise”.

Correlations cause relationships

Both economists and Sima aunty are always trying to find a relationship between two variables. Based on her client’s criteria, she starts making connections between personalities and desires. Data-driven thinkers assign a score between minus one and plus one, to see how closely two variables associate with each other. If none of the preferences match between two bio-datas, their score is 0, and there is no match. With more similarities, the match starts moving in the positive towards 1. This is the logic behind her exclaiming “you have a lot in common!”, to her clients. Once in a while, a correlation of -1 is at play. This happens, as they say, when opposites attract!

In season one, hard-to-quantify traits, like adjusting and flexible, dampened her success rates. And so, in season two, Sima aunty arrived in her clients' homes, and on our screens, with a limitation, “only 60-70 percent of the criteria will be met. Because everything is not possible”. She is aiming for what data scientists aim for: a good fit. An economist's jargon for it is an adjusted R square, aunty’s parlance is to see if the “stars can align”. The R square captures the strength of the match, the dependence of one variable on the other.

A 60-70 percent R square is a benchmark for a good predictive model for economists. Sima aunty also tries to convince her clients, like Shital Patel who wants more of a 90 percent match, that a person fulfilling 60-70 percent of her criteria will be a good fit. It is unlikely that Sima spends time with an intermediate statistics textbook or learns to model for her vocation. But with 16 years of experience, her brain engine has the innate power to compute the chances of finding her client their significant other.

Sima versus MIMA

What Sima aunty does in her head, Matrimony.com's Intelligent Matchmaking Algorithm (MIMA) does with artificial intelligence. Today, firms like Shaadi.com and BharatMatrimony, and dating apps such as Tinder, Bumble and Hinge operate on multiple models, making economists, statisticians and data-scientists a large part of the matchmaking market. Their datasets are larger than Sima’s ledger, and the platforms feed far more variables in their models, including sensitive data. To anyone who wants to rely on them, Sima aunty wishes, “good luck!”.

Dating apps are aware that they are overcrowded, which can often lead to option-paralysis. This is why Hinge offers liking a maximum of eight users, while Bumble and Tinder have daily swipe limits. Sima aunty is vocal about her “way of working” in the show, and it differs on this aspect. She does not present a catalogue of partners, and offers only two potential choices, working with a 50 percent chance, like A/B testing, a method used to test user experience. Her clients meet these matches sequentially, and never in parallel, as in the case of speed-dating.

Sima’s work is governed by the rules of the society. So are the predictive models of various matchmaking firms. These rules are set to the idea of a typical Hindu marriage, which is heteronormative and caste-based, where women’s age and skin colour matters still. Until 2020, even Shaadi.com required users to feed in their skin colour. A quick glance at the matrimony section of any newspaper will point to the demand for a homely, beautiful, fair, non-manglik bride. A logical method may lead people to their soulmates, or love may arrive out of the blue, but all efforts will remain meaningless, if the stars are not aligned.
Saurabh Modi is a public policy researcher. Views expressed are personal.
Surbhi Bhatia is an independent data journalist. Views expressed are personal.
first published: Sep 18, 2022 06:19 pm