Data_150_02_S21_Jack_Hayes

Assignment 3

Word Count: ~1400

Introduction

Gender inequalities and inequities have been affecting the world seemingly since the beginning of time. Especially with a recent surge in social justice, the battle against gender perspectives around the globe has gained more and more attention. However, many countries, like Japan, are still struggling to combat gender perspectives. One of the main sources of these gender-related issues in Japanese society is not as apparent as one would think, however. Data concerning the reinforcement of gender perspectives can be seen in almost every natural disaster in Japan’s recent history. For example, women have a significantly higher mortality rate than men in Japan and have to constantly voice their grievances about being unheard in their communities. This problem is extremely prevalent during the almost never-ending process of reconstruction in Japan. For example, the Great East Japan Earthquake occurred in 2011, and the social discord caused by the earthquake is still felt today, as reconstruction continues. The social cohesion of a population leads to a better educated, economically flourishing, happy society, but social cohesion cannot be achieved without the recognition and solving of gender inequities.

The lack of administrative participation is an inherent problem in the fight against these Japanese social issues. Whether it’s informing the public of said gender perspectives, or involving organizations like gender equality centers in disaster management systems, the high-ranking individuals within Japan’s government have a big role in this complex, nonlinear process. This leads to a broad research question: What policies and ideas can Japan’s authorities implement within their disaster management systems to combat gender perspectives and improve community resilience as a whole? To understand this evaluative and explanatory inquiry as a causal and predictive puzzle, one must understand how to fight gender perspectives and improve community resilience through data. To understand how to fight gender perspectives and improve community resilience through data, one must understand how the data is gathered in the first place. With cell phones becoming an integral part of many peoples’ lives, they have become an incredibly important tool in data collection. But how exactly are these handheld devices used to track and record migratory and behavioral activity across the globe? And how do they help us better understand social issues after a disaster strikes?

Call Detail Records

Using call detail records, or CDRs, is the other big way of collecting large geospatial datasets opposed to satellite-collected data. CDR refers to the mobile phone call detail records that track individuals. As one might guess, this is incredibly useful when inspecting the behavioral activities and patterns of people after a crisis. CDR analysis is an efficient method for tracking large numbers of people in a meaningful way without interview bias, but there are some inconsistencies in the collection of CDR data. Unfortunately, CDR data are almost never representative of an entire population, because only people who can afford and use SIM cards can be tracked. Especially in less developed countries, there can be a big problem of underrepresentation. Despite these negatives, CDR data is extremely useful when used as a complement to traditional methods. When looking at post-disaster resilience within a population, CDR-based analyses are paramount, as understanding the behavioral patterns of a population leads to more complete methodologies concerning the improvement of disaster response(Flowminder Foundation, 2019).

When tracking and analysing behavioral activity with CDR data, there are two main metrics: radius of gyration and entropy. The latter looks at the predictability of an individual’s path or course, whereas the former measures the linear space occupied by a person. Both of these metrics share four principal steps. The first is to calculate a mobility metric, typically weekly, in the dataset. The second is to create a benchmark value that defines an individual’s “normal” behavior by calculating a mean value over the weeks of the pre-disaster period. Thirdly, beginning on the first day of the disaster, a rolling mean value of the metric is gathered each week, usually for four weeks. Lastly, the resettling date is calculated by comparing the four week rolling value to the benchmark value(Flowminder Foundation, 2019).

Radius of Gyration

The radius of gyration is normally defined as “the standard deviation of the distances between each visited location and the centroid, over a specified time period”(Flowminder Foundation, 2019). The typical equation for this is: . The length of the time period is n, while rc is the centroid, calculated as the mean location based on the frequency of visits. Since standard deviations are skewed by unusually high values, long-distance trips are heavily weighted in this method. Additionally, if a small sample is being analyzed, using the absolute deviation is more accurate than the standard deviation. When plugging in the variables to this equation, one must find the mean distance from “home” every week as the centroid, calculate the absolute value of the difference between every daily location and its corresponding centroid, and calculate the average of the weekly absolute differences. This is the methodology used to find the true radius of gyration, which assumes all locations lie along a one-dimensional line. For inspecting changes in pre- and post-disaster periods, however, a reduction in complexity is needed. A team of researchers in a recent 2019 study on estimating disaster resilience did just that. By analyzing migration behaviors from the 2010 Haiti Earthquake, 2016 Hurricane Matthew, and 2011 Nepal Earthquake, the team observed that many relocation displacements occur over distances under a few kilometres. Because of this, the researchers developed a modified RoG that is more sensitive to short trips. Taking the logarithm of the modulus of the difference between daily locations and their centroids and calculating the mean of the weekly logarithmic differences ensures that short-distance travel is weighted heavier. This logarithmic radius of gyration (LRoG) is incredibly beneficial for assessing behavioral and migratory patterns after a disaster. The team found that using their modified radius of gyration and entropy metrics(which will be explained later), the resettlement rates, or time it takes for an individual to find “home”, in all three of the natural disasters investigated could be modeled as exponential decay curves. A better understanding of these migratory activities leads to improved disaster response planning for the future, a greater analysis of the social effects of a crisis, and finer data in general for conducting more studies relating to disaster resilience(Flowminder Foundation, 2019).

Entropy

In general, entropy quantifies the predictability of an individual’s movements, but there are several definitions of entropy depending on certain sensitive behaviors specific to certain events. True entropy looks at the probabilities of all subsequences within a time series and accounts for a relative frequency and order of visited locations. Temporal-uncorrelated entropy only accounts for a relative frequency and random entropy only depends on the number of distinct locations visited.

Like radii of gyration, however, these types of entropy can be modified to benefit the analysis of post-disaster migration. The team in the previously mentioned 2019 study also created an entropy-based metric sensitive to changes in location, but independent of distance. This new “step entropy” is based on Boltzmann’s entropy in statistical thermodynamics. In this metric, consecutive visits to the same location are reduced to a singular visit, and the locations visited are grouped as a sequence. The entropy is then defined as the logarithm of the distinct permutations of the total numbers of visits in said sequence and the total number of times a specific location occurs in the sequence.

To sum it up, this metric is simply a measure of disorder. The reason this type of entropy is so important when it comes to disaster management is because of how fast it can display migratory patterns when individuals are highly mobile. This is incredibly useful after an earthquake, for example, when individuals are constantly moving from temporary shelter to temporary shelter in search for “home”(Flowminder Foundation, 2019).

https://www.researchgate.net/profile/Xin-Lu-31/publication/227343668/figure/fig2/AS:667201602330626@1536084666386/Trajectory-analysis-of-mobile-phone-users-who-were-present-in-PaP-on-the-day-of-the.png Example of a radius of gyration and entropy graph

Conclusion

When considering the analyses of data concerning large social issues, it is crucial to ask how that data was gathered. Altering equations of metrics to best fit the scenarios of certain areas of research can lead to clearer, more accurate conclusions. Relating to Japan’s disaster management organization specifically, these metrics can be implemented to grasp a better understanding of behavioral patterns relating to social issues for pre- and post-disaster times. For example, this methodology can find data that shows what communities are left the most due to social discord, and what communities are joined because of their social open-mindedness. Using a standard radius of gyration metric along with a temporal-uncorrelated entropy metric alongside satellite-gathered data and surveys will exponentially increase a nation’s understanding of social effects during a crisis. This newfound understanding will help countries like Japan in their development processes, as crippling social unfreedoms like gender perspectives will be easier to reform, especially if the newfound understanding is publicly displayed. Although a step forward, there are still questions needed to be answered regarding Japan’s way forward against gender perspectives. Even with a great understanding and analysis of social behaviors before, during, and after a disaster, how can this synthesized data realistically be used to help actively combat these evils?

Reference

Li, T., Dejby, J., Albert, M., Bengtsson, L., & Lefebvre, V. (2019, August 06). Estimating the resilience to natural disasters by using call detail records to analyse the mobility of internally displaced persons. Retrieved from https://www.flowminder.org/resources/publications/estimating-the-resilience-to-natural-disasters-by-using-call-detail-records-to-analyse-the-mobility-of-internally-displaced-persons/