EDA on DETER Ground Truths.md

Initial Exploratory Data Analysis (EDA) of the available ground truth

Analysis of Classnames with their English translations

Classname	Classname (English)	Count of incidents
CICATRIZ_DE_QUEIMADA	scar burned	34886
CORTE_SELETIVO	selective cutting	1
CS_DESORDENADO	cs cluttered	1543
CS_GEOMETRICO	cs geometrico	1415
DEGRADACAO	degradation	19490
DESMATAMENTO_CR	deforestation cr	119470
DESMATAMENTO_VEG	veg deforestation	5957
MINERACAO	mining	3651
Grand Total		186413

Count of UF

State Name	Code	Count of incidents	SUM of AREAMUNKM	AVERAGE of AREAMUNKM	% of TOTAL AREA
Acre	AC	13086	1,882.82	0.1439	7.97%
Amazonas	AM	21313	6,804.17	0.3192	1.37%
Amapá	AP	362	153.87	0.4251	0.25%
Maranhão	MA	5302	6,109.59	1.1523	1.61%
Mato Grosso	MT	39629	42,840.92	1.0810	4.39%
Pará	PA	72627	36,864.50	0.5076	5.83%
Rondônia	RO	26169	6,566.90	0.2509	11.01%
Roraima	RR	6570	6,065.89	0.9233	2.93%
Tocantins	TO	1353	2,205.39	1.6300	0.49%
Grand Total		186411	109,494.06	0.5874

Visualizing areas of interest

Link to reactive image

Large outliers skew the data

Alfa-distribution after excluding outliers (AREAMUNKM <10 )

Timeseries of AREAMUNKM over time, Large spikes around Aug-Sept