Poverty Prediction Challenge

Goal of the Competition

Accurate poverty measurement is essential for directing global development efforts and informing evidence-based policies for poverty reduction and equity enhancement, yet many countries lack recent data due to the high costs and complexity of collecting comparable comprehensive household expenditure surveys.

This challenge simulated a common real-world scenario faced by economists, who are tasked with producing up-to-date poverty measurements and additional welfare indicators, even in cases where fully detailed recent information on household expenditure is unavailable. The goal was to develop survey-to-survey imputation models that predicted both poverty rates and per capita household consumption from anonymized historical survey data.

Performance was evaluated according to a weighted average of the household-level prediction error and the distribution-level prediction error:

• 90% of the weighted average was computed as the weighted mean absolute percentage error (w-MAPE) between predicted poverty rates and the actual rates at 19 specific consumption thresholds ranging from $3.17 to $27.37
• 10% consisted of a mean absolute percentage error between predicted household-level per capita consumption and actual per capita consumption (measured in 2017 USD PPP)

See the World Bank's results page of this competition here.

What's in this Repository

This repository contains code from winning competitors in the Poverty Prediction DrivenData challenge. Code for all winning solutions are open source under the MIT License.

Winning code for other DrivenData competitions is available in the competition-winners repository.

Winning Submissions

Place	Team or User	Public Score	Private Score	Summary of Model
1	dwivedy045	5.4466	5.7545	LightGBM pipeline with grouped cross-validation (GroupKFold by survey), categorical handling, and quantile-mapped inference calibration for household consumption and poverty-rate prediction.
2	Khartoum	12.0761	7.7052	LightGBM with leave-one-survey-out cross-validation, anti-leakage survey-specific mean ratio features, top-75% feature selection by importance, and weighted quantile calibration for poverty rate distribution matching.
3	selman	8.1626	8.2382	4-model gradient boosting ensemble (LightGBM ×2, XGBoost, CatBoost) with per-capita feature engineering, test-time augmentation over utility expense perturbations, P40-focused survey matching, and per-survey calibration via differential evolution optimization.

Competition Leaderboard

Rank	Participant	Private wS-wMAPE	Private wMAPE
🥇 1	dwivedy045 🏆	5.7545	2.6177
🥈 2	shyboy	7.1904	3.9436
🥉 3	Khartoum 🏆	7.7052	4.5917
4	Daiz	7.7098	4.5757
5	ganontha	8.1882	5.1818
6	selman 🏆	8.2382	4.9622
7	javidkamal15	8.2692	4.9556
8	pavnoval	8.2782	5.0912
9	chidubem	8.4354	4.8644
10	HeroVoltsy	8.6809	4.7584
11	jekiwantaufik	8.8876	5.0395
12	charles625	9.1140	6.2603
13	oc166	9.2707	5.4388
14	kasrsf	9.2846	6.1969
15	govinuts	9.5113	5.4497
16	jvillines	10.1848	6.9231
17	mgoulart	10.2600	7.0326
18	BoundaryLab	10.3068	6.9856
19	vbhv3773	10.4797	6.9985
20	juliop_	10.4991	7.5255
21	eferraz	10.5644	7.6719
22	rokket9000	10.7479	7.4929
23	kefthyme	10.7736	7.7727
24	Nicole_Tiokhin	10.8779	7.5150
25	m39lee	10.8856	7.5195
26	ananya_s	10.9986	7.5196
27	hafidh24	11.0050	7.6435
28	tiesp	11.0223	7.5210
29	太上老君	11.0546	8.0528
30	wilfred3235	11.2403	8.2777
31	Lucian41	11.2548	7.9212
32	rudecia	11.4028	7.9651
33	_0037	11.5254	7.5210
34	TheIsoLab	11.6938	8.0978
35	lazarosgogos	11.8047	8.8482
36	azmiy	11.8370	8.8265
37	mithos	11.9437	8.9990
38	bbeum	12.0760	9.1018
39	Ayush_killer	12.1643	8.8979
40	pecama4267	12.1961	8.8370
41	dmitrysarov	12.4571	9.2217
42	NxGTR	12.4870	9.4115
43	YESSEth (team)	12.7809	9.6397
44	Yuxinnn	12.9049	9.4346
45	Merch1000	13.0533	10.2115
46	NataliaTAmv	13.0591	9.6643
47	EconAI (team)	13.1496	9.8068
48	limzero	13.1733	10.2743
49	53un	13.2333	9.8654
50	No Comeback Crew (team)	13.3083	9.7532
51	barata.lade	13.3276	10.1708
52	phucdkbk	13.3573	10.4526
53	isidorat	13.3869	9.3881
54	PaulMcBride	13.4013	10.5148
55	Kosmas7	13.4500	10.2108
56	jackson5	13.4801	10.5671
57	Smasko	13.5287	10.2108
58	Stealth (team)	13.5598	10.4494
59	sachin__09	13.5902	10.4205
60	Chinchpokli	13.7670	10.4223
61	Vincent Schuler (team)	13.7782	10.9191
62	sumatorikki	13.7953	10.9119
63	qgoens	13.8934	10.5628
64	kar_len	13.9865	10.8769
65	KallK	14.0341	10.8172
66	ssimp10032	14.0453	10.5108
67	emmettsexton	14.0570	9.8790
68	ocitalis	14.0791	11.0760
69	structAS	14.1863	11.0981
70	deactivated-e9b8d89	14.1911	11.3454
71	dami	14.3137	10.8556
72	meenalmurugesan	14.4510	11.1949
73	yummyfe	14.4712	10.2904
74	shenjianmantou (team)	14.4881	11.5891
75	Athirajan	14.5131	11.1949
76	DataForGood (team)	14.5520	11.2561
77	koni-team (team)	14.5538	11.2346
78	Surgis12	14.5605	11.5682
79	Helio	14.6824	11.3642
80	shimaa	14.7513	11.2211
81	jakey	14.7643	11.4462
82	mugesh9085	14.7742	11.4561
83	jil300	14.7742	11.4561
84	kiochan	14.7742	11.4561
85	Himfs	14.7742	11.4561
86	sheldon9789	14.7785	11.4603
87	mashy	14.8077	11.4896
88	tika	14.8096	11.4915
89	Watson9789	14.8106	11.4924
90	ragnar9085	14.8176	11.4995
91	lothbrok	14.8277	11.5095
92	Buhari2612	14.8367	11.5185
93	CJ343	14.8440	11.9419
94	Sherlock281	14.8701	11.5519
95	Rmugesh	14.8978	11.5797
96	mugesh2000	14.8981	11.5799
97	AmineSamoudi	14.9007	12.0550
98	magist6	14.9299	11.9307
99	MAT-U	14.9796	12.1361
100	dishantsaini55	15.0494	12.1644

🏆 = prize winner  |  (team) = multi-person team entry