Exploratory Data Analysis on Brazil forest fire using Plotly

TINU ROHITH D
Analytics Vidhya
Published in
4 min readNov 20, 2019

EDA(Exploratory Data Analysis): It is a term for certain kinds of initial analysis and insights done with data sets, usually early on in an analytical process. “It's kind of a story telling through visualization of graphs”.

To get a “feel” for the data, it is not enough for a data scientist to know what is in the data; also a data scientist must know what is not in the data, and the only way to do that is to draw on our own human pattern-recognition and comparative abilities in the context of a series of judicious graphical techniques applied to the data.

Plotly: Its an interactive analytics/statistic dashboard, that brings the insights into better visualizations for the user to view.

Every Data Scientist and an aspiring Data Scientist must be aware of interactive analytics/statistics dashboard visualization like plotly, tableau, microsoft power BI and so on.

Let's get into eda analysis on brazil forest fire: Here, the dataset contains information on number of forest fire in Brazil divided by states for a period between 1998 to 2017. To download the dataset click on below link https://www.kaggle.com/gustavomodelli/forest-fires-in-brazil

Links for viewing interactive plots are pasted at the end of blog:

Occurence of forest fire reported on top 10 states for the period between 1998–2017 :

The graph shows the occurrence of forest fire reported in top 10 states for a period of 20 years. Over the years (1998–2017) around 90k times there has been forest fire occurrence in Mato Grosso. 30k times there has been occurance of forest fire in Amazonas. Pariba and Sao Paulo has a forest fires count of 50k.

Code:

Total number of forest fire across each month for top 10 states:

The below chart explains the count of forest fire reported on top 10 states across each month. Marco, Abril, Fevereiro, Maio(March,April,February,May) have low count of forest fire occurrence compared to remaining months. July, August, November and October has a a high report of forest fire over subsequent years.

Code:

Mapping locations of Brazil States:

The below chart shows the location of each states of Brazil that has reported with forest fire.

Code:

Report of forest fire across months over the period 1998–2017:

The below chart explains, report of forest fire over months. August, July, October and November has highest number of forest fire occurrence around 80k to 90k times over the past 20 years.

Contribution of forest fire across years for top 10 states:

The below chart provides count of forest fire occurrence for every year from 1998–2017 across top 10 states that has reported with highest number of forest fire.

Code:

Occurance of forest fire over the period of years:

The below chart explains the forest fires over the years 1998–2017. There has be a subsequent increase of forest fire between 2001 to 2003 and subsequent drop between 2003 to 2008.

There has been a gradual increase from 2008 to present.

Code:

Percentage distribution of forest fires by across each state and month:

The below chart explains the percentage distribution of forest fires reports across months. On an average there are 12% to 13% of forest fire occurrence during the month of July, October August and November. 4% to 5% of occurence of forest fire across March to May.

The below chart explains the percentage distribution of forest fires reports across states. On an average there are 7% to 8% of forest fire occurrence across Sao Paulo, Mato Grosso and Bahia. 0% to 1% of occurence of forest fire (least chances of reports on forest fire) across Alagoas, Distrito Federal and Sergipe.

Code:

Conclusion:

2001 to 2003 has reported highest number of forest fire. São Paulo and Mato Grosso states have high report of forest fires. Alagoas, Distrito Federal and Sergipe have least contribution of forest fire occurrence over the past years 1998–2017. Month of August to November enquires high number of forest fires every subsequent years .

Link to view the interactive graphs and code:

--

--