Leveraging Satellite Imagery to Analyze Large-Scale Economic Trends

Poverty is still one of the factors that have been affecting the lives of humans over hundreds of decades in the world but still, it is not entirely eradicated. One of the main reasons is the lack of identification of the economical pattern in a particular area that has been affected by poverty over a long period which has not been noted by any NGOs or government. So, that is the main reason why poverty eradication is still the first goal of UN sustainable development goals. [1] Understanding economical patterns over a region can lead us in providing informed policy-making, targeted NGO, and government-aided efforts. If we could track them down in an easy method at the earlier stage or in the advanced method and predict this fatal enigma on the region, the poverty can surely prevent or at least save lives. The above dilemma leads us to the essential method of tracking the reliable and timely measurements of economic activities, which are key for understanding economic development and designing government policies. But still many countries, especially developing countries, lack reliable data. Though data is available, the lack of quality in the data remains a huge challenge. In this paper, we propose a low-cost yet efficient approach to predict the economical pattern in a rural region from satellite imagery, which is globally available, in the meantime, satellite images are continuously updating to the changeable environmental conditions. Since the satellite image is way more advantageous than traditional methods, they play a huge role in the model. Traditional methods like surveys are expensive to conduct and include a lot of manpower. [2]This contributes to infer socioeconomic indicators from large-scale, remotely-sensed data. The available dataset is used to extract the luminous intensity from the night-time satellite images and added along with the other features of the particular region that determines the socio-economic conditions. Then with machine learning algorithms, the range of economic class is built and categorized accordingly.