Migration patterns

ISB’s SRITNE (Srini Raju Centre for IT and the Networked Economy) is a re­search canter that fo­cuses on the busi­ness and so­ci­etal value of IT. Gramener col­lab­or­ates with SRITNE to de­velop and pro­mote visu­al ana­lyt­ics, and to help foster a cul­ture of open data with­in the com­munity.

As part of this col­lab­or­a­tion, we jointly presen­ted ‘Visualisation of Migration Patterns in India’ at the Bangalore Open Data Camp. Access to the source data­set for this ana­lys­is was provided by ISB and Visualisations were done primar­ily on the Gramener Visualisation Server. MS Excel and R were used for ex­plor­at­ory ana­lys­is. The aim of this ex­er­cise was to take an out­side, ana­lyt­ics view of the Migration Patterns to ex­plore al­tern­ate pos­sib­il­it­ies of rep­res­ent­a­tion and visu­al­isa­tion, rather than from the lens of a Demographics Analysis Expert.

The Indian NSSO (National Sample Survey Office) had con­duc­ted the 64th round sur­vey on ‘Employment & Unemployment and Migration Particulars’dur­ing July ’07 to June ’08 cov­er­ing 1,25,578 house­holds and 5,72,254 per­sons.

Of this sample, ~30% were found to be mi­grants, i.e. those whose last usu­al place of res­id­ence (UPR) was dif­fer­ent from the present place of enu­mer­a­tion. In this sur­vey, the usu­al place of res­id­ence of a per­son was defined as a village/town where the per­son had stayed con­tinu­ously for six months or more. Amongst the mi­grants, a ma­jor­ity were found to be mov­ing with­in the state (85%) as op­posed to those mov­ing across states (15%). Women formed a size­able ma­jor­ity of this mi­grant pop­u­la­tion.

Intra-state mi­gra­tion pat­terns

image

The map on the left has the intra-state mi­gra­tion pat­tern (ex­clud­ing inter-state num­bers) show­ing the ab­so­lute num­ber of mi­grants mov­ing with­in each state/UT. Green in­dic­ates higher mi­gra­tion and red is the op­pos­ite. Based on this map, the 5 most pop­u­lous states in India ac­count for the top 5 intra-state move­ments, ex­cept for Bihar which comes a close 6th. If we res­cale the num­bers by tak­ing mi­grants as a per­cent of the state/UT’s sur­vey size, as shown in the right map, the res­ults change com­pletely. The top 5 states with highest per­cent churn are Andhra Pradesh, Himachal Pradesh, Kerala, Gujarat and Andaman & Nicobar Islands.

Inter-state mi­gra­tion pat­terns

image

If we now look at the Inter-state mi­gra­tion pat­tern (ex­clud­ing within-state move­ments) by plot­ting the Net Inflow of mi­grants in­to each state/UT (left-hand-side map), the states with highest net out­flow of mi­grants are Uttar Pradesh and Bihar, while those with highest net in­flow are Maharashtra and Delhi. If we res­cale the num­bers, as a per­cent of the state/UT’s sur­vey sample, the story changes, yet again. All the Union Territories in India have the highest Net per­cent Inflow, with Chandigarh show­ing the highest value at 41%.

Inter-state mi­gra­tion Heat-map

state-migration-heatmap

In or­der to get a sense of ex­change of mi­grants hap­pen­ing between the states, we plot­ted the num­bers on a heat-map. The y-axis of the heat­map has ‘From-State’ while ‘To-State’ is on the x-axis. The height of each heat-map box is pro­por­tion­al to the net out­flow from the contributor-state, while the width of each box is pro­por­tion­al to the net in­flow in­to the recipient-state. The col­our is rep­res­ent­at­ive of the num­ber of people mov­ing between the states – dark­er the box, more the num­ber of people.

As can be seen, the top des­tin­a­tions for people leav­ing UP are Delhi, Maharashtra and Uttaranchal re­spect­ively. For Bihar and Rajasthan, the top des­tin­a­tions are high­lighted ac­cord­ingly. What is more in­ter­est­ing is the pat­tern of top des­tin­a­tions for each of the states. A clear trend is the con­sist­ent pref­er­ence of people across re­gions to mi­grate in­to states with geo­graph­ic­al prox­im­ity. The sur­vey had also covered a set of in­ter­na­tion­al in-migrants, where­in Bangladesh the top con­trib­ut­ing coun­try has a size­able pro­por­tion of its mi­grants mov­ing to West Bengal.

Migration across Rural-Urban areas

image

When mi­gra­tion was viewed from the per­spect­ive of move­ment across Rural – Urban areas, a sur­pris­ing trend found was the ex­tent move­ment with­in Rural Areas – more than half of mi­gra­tion in India hap­pens among­st the Rural re­gions. About 40% of mi­gra­tion is to­wards Urban areas. A contra-trend no­ticed here was for the Union Territories and North-Eastern States – over 70% of mi­gra­tion in these areas is to­wards the Urban re­gions, un­like the rest of India.

Reasons for mi­gra­tion

migration-reason-age-gender

When an ana­lys­is of Reasons for Migration was done at the Country level, some key trends were ob­served. Women, who form a size­able ma­jor­ity of the mi­grants primar­ily mi­grate on ac­count of ‘Marriage’ and their typ­ic­al age at mar­riage is between 15 and 24. For men, the key reas­on for mi­gra­tion is ‘Employment-related’ and this primar­ily hap­pens in the age band of 18 to 40. Consequently, mi­gra­tion due to ‘Movement of Parent/Earning mem­ber’ forms an­other key reas­on. ‘Education’ is also found to be a driver of mi­gra­tion and this typ­ic­ally hap­pens for men and wo­men un­til the age of around 23 years.

When we looked at the Reasons for Migration vis-à-vis States, a few in­ter­est­ing pat­terns showed up. People in Tripura mi­grate mostly due to Forced Reasons/Disasters, where­as UP wit­nesses Marriage-related move­ment. Kerala and West Bengal wit­ness mi­gra­tion be­cause of Housing re­lated reas­ons, where­as a lot of people in the scen­ic state of Himachal Pradesh mi­grate for post-retirement life.

migration-status-reason

It is evid­ent from the above heat­map that a ma­jor­ity of the wo­men who mi­grate for mar­riage, end up do­ing Domestic du­ties, while men who move for em­ploy­ment end up as Wage employees/labourers.

migration-reason-year-gender

The sur­vey sample had a good mix of people who had mi­grated over the years, dat­ing as far back as the 1930s. When we ana­lysed the pat­tern of evol­u­tion of mi­gra­tion reas­ons, in­ter­est­ing trends emerged. Until Independence, mi­gra­tion was sub­dued and was re­stric­ted only to the wo­men get­ting mar­ried. Post-independence, mi­gra­tion num­bers have stead­ily in­creased over the next 60 years. After 1970s, in­creas­ingly more people star­ted mov­ing for Employment-related reas­ons. This was also ac­com­pan­ied with mi­gra­tion of the de­pend­ent fam­il­ies. It has been only af­ter the 1990s that people move in sig­ni­fic­antly lar­ger num­bers and for reas­ons such as Business, Education, Housing, Post-retirement, Healthcare – more in­line with the Indian Economic Development story over the past 60 years!

Karnataka ground water quality

We took Karnataka’s ground wa­ter qual­ity data from a 2004 Karnataka Rural Water Supply and Sanitation Agency (KRWSSA) re­port (via IndiaWaterPortal), and tried to see if there were any pat­terns.

The ex­ec­ut­ive sum­mary of the re­port shows the num­ber of vil­lages in each dis­trict af­fected by prob­lems of ex­cess ni­trate, iron, flu­or­ide or total dis­solved salts. We plot­ted those on the dis­trict map.

karnataka-water-quality-2004

The in­ter­est­ing pat­tern is that emerges is that the places that have ex­cess ni­trate (NO3) con­cen­tra­tion are the ones that have ex­cess iron (Fe) con­cen­tra­tion as well. The places that have ex­cess flu­or­ide (F) con­cen­tra­tion are the ones that have ex­cess salts (TDS).

karnataka-water-quality-correlation

The cor­rel­a­tion scat­ter­plot along­side fur­ther demon­strates this point. There is a fairly good cor­rel­a­tion between Fe and NO3, and between TDS and F. There’s little cor­rel­a­tion across these groups how­ever.

(From a glance at the scat­ter­plots, though, it be­comes im­me­di­ately ob­vi­ous that this is based on too few data points.)

We can also read­ily see that the bulk of the wa­ter qual­ity is­sues are in in­teri­or Karnataka. The coastal areas are re­l­at­ively fine.

Geographic visu­al­isa­tions are an ex­tremely power­ful way of in­fer­ring pat­terns when the un­der­ly­ing data is geo­graph­ic in nature. At Gramener, we use our visu­al­isa­tion server to auto­mat­ic­ally cre­ate graphs such as these based on an un­der­ly­ing data source.

We have provided a sample of our tool at http://gramener.com/indiamap. You can en­ter your own data and see how it shows up on any dis­trict or state map. Happy map­ping!

Visualising Indian Elections

When the Election Commission res­ults for the 2011 as­sembly elec­tions came out, we thought we’d take a look at the res­ults for Tamil Nadu.

The easi­est way to see geo­graph­ic data is, of course, on a map. Since the res­ults are an­nounced constituency-wise, it makes lo­gic­al sense to plot each con­stitu­ency on the map.

However, map-based visu­al­isa­tions suf­fer from one prob­lem: the area of a geo­graph­ic re­gion is not al­ways pro­por­tion­al to its im­port­ance. In elec­tions, each con­stitu­ency equal weight­age: one seat. But the areas can vary con­sid­er­ably. Chennai, for in­stance, is a tiny dot on the map, but yet is split in­to three dif­fer­ent con­stitu­en­cies.

A bet­ter way is to take an ap­prox­im­a­tion of the map. We can plot each con­stitu­ency as a block, and po­s­i­tion it roughly at its geo­graph­ic loc­a­tion, and roughly pre­serve ad­ja­cency. This gives a reas­on­ably good geo­graph­ic pic­ture, while show­ing re­l­at­ive im­port­ance ac­cur­ately.

For ex­ample, here is a world map: the area is pro­por­tion­al to each country’s pop­u­la­tion.

world-population

We did the same for Tamil Nadu. Below is an in­ter­act­ive map that shows voter turnout. Hovering over each cell will show the num­ber of voters that turned out in that con­stitu­ency. Black shows higher votes, white shows few­er voters.

Voter turnout

Men turnout

Women turnout

The num­ber of the re­gistered voters ranges from 140,000 (Kilvelur) to 360,000 (Shozhinganallur). Each box rep­res­ents one con­stitu­ency, with a po­s­i­tion ap­prox­im­at­ing its geo­graph­ic loc­a­tion. This is a reas­on­ably good proxy for a con­stitu­ency dens­ity map.

Now, let’s take a look at voter turnout.

Turnout%

Turnout% men

Turnout% women

The voter turnout was pretty high, at 78%. Not too dif­fer­ent between the men (77.7%) and wo­men (78.5%), but quite dif­fer­ent across con­stitu­en­cies. Palacodu, Kulithalai and Veerapandi led the pack with turnouts of 87%, 89% and 89% while Chennai Harbour, Killiyoor and Colachal bot­tom at 64% each. The lowest turnouts are con­cen­trated around Chennai, Coimbatore and Kanyakumari – in­ter­est­ingly these are the more af­flu­ent areas.

% votes: ADMK

% votes: DMDK

% votes: DMK

Clearly, ADMK has swept the elec­tions. What’s in­ter­est­ing is that they won ab­so­lute ma­jor­it­ies in a num­ber of areas (high­lighted by the dark­er shades of green), un­like DMK or MDMK, who mostly won sim­ple ma­jor­it­ies.

% margin: ADMK

% margin: DMDK

% margin: DMK

A num­ber of ADMK vic­tor­ies were by a large mar­gin, but so were some of the DMDK and DMK vic­tor­ies. AKT Raja of DMDK won Thirupurankundram with a 30% mar­gin, for ex­ample. Former Chief Minister M. Karunanidhi (DMK) won with a com­fort­able 29% mar­gin at Thiruvarur as well. Vijaykant (DMDK) man­aged an 18% mar­gin at Rishivandiyam. It is in­ter­est­ing that 92% wo­men voted at Rishivandiyam – sig­ni­fic­antly higher than any­where else in the state.