8
analysis. The model with the best performance (the logistic
regression model) was selected such that it can be deployed
to classify new and unseen text data (narratives) in the
MSHA dataset. The logistic regression model was utilized
to classify/predict the ground-fall incidents in the entire
MSHA dataset between 1983 and 2021. The result of the
logistic regression model is shown in the Dashboard section
in Figure 9.
The authors used the general outlines in Figure 6 to
demonstrate visualization of ground-fall trends in U.S.
coal mines between 1983 and 2021, named in this paper
as the Dashboard. The Plotly Dash using Python was used
to generate an interactive and dynamic Dashboard (Plotly
Technologies Inc., 2015). Other programming languages,
such as R can be used to achieve the same goal. Figure 6
shows the general structure of a Dash app. In the imports
part, the required packages are imported. The app instan-
tiation is a straightforward way to create the app. The app
layout is used to lay out various containers, dropdown
menus, and figures. The callback functions link different
elements of the layout and make the Dashboard interactive
and dynamic.
A benefit example of the Dashboard is the potential
to identify areas where additional research is needed and
where innovative solutions may need to be developed
to reduce these potentially severe occupational hazards.
Hence, the Dashboard would be a useful tool to identify
ground control related health and safety gaps to areas that
experience more injuries/fatalities. Additionally, the use of
these tools may complement surveillance statistics efforts to
track trends such as reduction in ground fall incidents in
U.S. coal mines.
Figure 7 shows the general layout for the prototype
Dashboard that was developed for internal use only. This
Dashboard, developed to identify ground fall trends in the
U.S. coal mines, is composed of four tabs: Project Info,
Mining Operations, Mining Methods, and Ground-fall
Classification. The Project Info tab explains the goal and the
outcome of the Dashboard. It also shows three examples of
ground-fall incidents (roof fall, rib fall, and outburst) only
one example is shown in Figure 7.
The Mining Operations tab is divided into two panels
the left panel includes three dropdown menus and a slider
bar to filter the ground-fall incident database. In the first
dropdown menu, the user of the Dashboard can select the
mining operation which would be one of the following:
surface, underground, or both surface and underground.
In the second dropdown menu, the user can select the
degree of injury due to ground-fall incidents, examples of
the degree of injury are: fatal, days away from work only,
and no injury. Note that when the degree of injury is “no
injury,” that means a ground fall incident occurred with-
out injury such that the rock fall blocked the ventilation
or was above the anchorage horizon, which is why it had
to be reported to MSHA using the mine the MSHA Form
7000-1 (MSHA, 1986). In the third dropdown menu, the
user can filter the ground-fall incidents based on the state,
and the slider bar is used to filter the data based on the time
period of interest. The Dashboard is dynamic and interac-
tive such that the plots in the right panel of Figure 8 would
Figure 6. The general structure of a Dash app to generate a Dashboard
analysis. The model with the best performance (the logistic
regression model) was selected such that it can be deployed
to classify new and unseen text data (narratives) in the
MSHA dataset. The logistic regression model was utilized
to classify/predict the ground-fall incidents in the entire
MSHA dataset between 1983 and 2021. The result of the
logistic regression model is shown in the Dashboard section
in Figure 9.
The authors used the general outlines in Figure 6 to
demonstrate visualization of ground-fall trends in U.S.
coal mines between 1983 and 2021, named in this paper
as the Dashboard. The Plotly Dash using Python was used
to generate an interactive and dynamic Dashboard (Plotly
Technologies Inc., 2015). Other programming languages,
such as R can be used to achieve the same goal. Figure 6
shows the general structure of a Dash app. In the imports
part, the required packages are imported. The app instan-
tiation is a straightforward way to create the app. The app
layout is used to lay out various containers, dropdown
menus, and figures. The callback functions link different
elements of the layout and make the Dashboard interactive
and dynamic.
A benefit example of the Dashboard is the potential
to identify areas where additional research is needed and
where innovative solutions may need to be developed
to reduce these potentially severe occupational hazards.
Hence, the Dashboard would be a useful tool to identify
ground control related health and safety gaps to areas that
experience more injuries/fatalities. Additionally, the use of
these tools may complement surveillance statistics efforts to
track trends such as reduction in ground fall incidents in
U.S. coal mines.
Figure 7 shows the general layout for the prototype
Dashboard that was developed for internal use only. This
Dashboard, developed to identify ground fall trends in the
U.S. coal mines, is composed of four tabs: Project Info,
Mining Operations, Mining Methods, and Ground-fall
Classification. The Project Info tab explains the goal and the
outcome of the Dashboard. It also shows three examples of
ground-fall incidents (roof fall, rib fall, and outburst) only
one example is shown in Figure 7.
The Mining Operations tab is divided into two panels
the left panel includes three dropdown menus and a slider
bar to filter the ground-fall incident database. In the first
dropdown menu, the user of the Dashboard can select the
mining operation which would be one of the following:
surface, underground, or both surface and underground.
In the second dropdown menu, the user can select the
degree of injury due to ground-fall incidents, examples of
the degree of injury are: fatal, days away from work only,
and no injury. Note that when the degree of injury is “no
injury,” that means a ground fall incident occurred with-
out injury such that the rock fall blocked the ventilation
or was above the anchorage horizon, which is why it had
to be reported to MSHA using the mine the MSHA Form
7000-1 (MSHA, 1986). In the third dropdown menu, the
user can filter the ground-fall incidents based on the state,
and the slider bar is used to filter the data based on the time
period of interest. The Dashboard is dynamic and interac-
tive such that the plots in the right panel of Figure 8 would
Figure 6. The general structure of a Dash app to generate a Dashboard