Weather-driven multi-category infrastructure impact forecasting
A method, system, and computer program product for resource management are described. The method includes selecting trouble regions within the service area, generating clustered regions, and training a trouble forecast model for the trouble regions for each type of damage, the training for each trouble region using training data from every trouble region within the clustered region associated with the trouble region. The method also includes applying the trouble forecast model for each trouble region within the service area for each type of damage, determining a trouble forecast for the service area for each type of damage based on the trouble forecast for each of the trouble regions within the service area, and determining a job forecast for the service area based on the trouble forecast for the service area, wherein the managing resources is based on the job forecast for the service area.
1 . A non-transitory computer program product, the non-transitory computer program product comprising a computer readable storage medium having instructions embodied therewith, the instructions executable by at least one processor to perform a method comprising:
receiving a plurality of trouble regions within a service area;
determining clustered regions, each of the clustered regions including at least one of the trouble regions within the service area, the at least one of the trouble regions of each cluster region being clustered based on similar weather and similar weather-related damage at least one of the clustered regions including at least two of the trouble regions within the service area, the at least one of the plurality of trouble regions being smaller than the service area;
receiving historical weather data of any number of the plurality of trouble regions;
for each clustered region, clustering the historical weather data for each of the trouble regions of that particular clustered region to generate clustered historical weather data for that particular clustered region;
for each of the trouble regions:
training a plurality of trouble forecast models for each type of weather-related damage, each of the plurality of trouble forecast models being for a different type of weather-related damage, each of the plurality of trouble forecast models being trained using the clustered historical weather data for the clustered region that the particular trouble region is a member;
applying one or more of the trouble forecast models for the at least one troubled region using a new weather forecast for an area that includes the troubled region within the service area to obtain a trouble forecast for the particular trouble region within the service area; and
continuously updating the trouble forecast models with new weather data from one or more weather observation sources by applying data filters to filter out low confidence data from the new weather data, wherein the low confidence data comprises data that fails to meet data source-specific quality thresholds, transforming the new weather data to common weather features including at least one of relative humidity, station pressure, temperature, wind speed, wind direction, pressure, accumulated rain, or rain rate, calibrating training input features using statistical influence techniques and retraining the trouble forecast models using the calibrated new weather data;
determining a regional job forecast for a geographic service area based on the trouble forecast for the geographic service area according to a trouble-to-job mapping, wherein managing resources is based on the regional job forecast for the service area; and
providing the regional job forecast for the geographic service area in order to initiate resource facilitate management of resources by relocating equipment and personnel within the service area according to the regional job forecast, in order to minimize an impact of the trouble forecast.
2 . The non-transitory computer program product of claim 1 , wherein the impact comprises an amount of time required to resolve weather-related damage to the resources.
3 . The non-transitory computer program product of claim 1 , wherein the impact comprises an amount of time required to resolve a weather-related interruption to service in the service area.
4 . The non-transitory computer program product of claim 1 , wherein the training the plurality of trouble forecast models is further based on trouble history records.
5 . The non-transitory computer program product of claim 1 further comprising collecting historical data from one of a National Oceanic and Atmospheric Administration (NOAA) Metar station or a NOAA Mesonet station.
6 . The non-transitory computer program product of claim 1 , wherein determining clustered regions further comprises generating cluster regions based on similarity of the trouble regions in at least one of job type or distance.
7 . The non-transitory computer program product of claim 1 , wherein determining the regional job forecast for the service area based on the service area comprises determining a forecasting horizon for each trouble region and each type of weather-related damage.
8 . The non-transitory computer program product of claim 7 further comprises, dividing the forecasting horizon into a plurality of prediction periods.
9 . The non-transitory computer program product of claim 8 , wherein determining the regional job forecast for the service area based on the service area further comprises computing, for each trouble region, trouble type and the plurality of prediction periods, a trouble count forecast based on a probability distribution.
10 . A system for managing resources based on weather-related damage in a service area, the system comprising:
at least one processor; and
memory including instructions executable by the at least one processor to configure the at least one processor to:
receiving a plurality of trouble regions within a service area;
determining clustered regions, each of the clustered regions including at least one of the trouble regions within the service area, the at least one of the trouble region of each cluster region being clustered based on similar weather and similar weather-related damage at least one of the clustered regions including at least two of the trouble regions within the service area the at least one of the plurality of trouble regions being smaller than the service area;
receiving historical weather data of any number of the plurality of trouble regions;
for each clustered region, clustering the historical weather data for each of the trouble regions of that particular clustered region to generate clustered historical weather data for that particular clustered region;
for each of the trouble regions:
training a plurality of trouble forecast models for each type of weather-related damage, each of the plurality of trouble forecast models being for a different type of weather-related damage, each of the plurality of trouble forecast models being trained using the clustered historical weather data for the clustered region that the particular trouble region is a member;
applying one or more of the trouble forecast models for the at least one troubled region using a new weather forecast for an area that includes the troubled region within the service area to obtain a trouble forecast for the particular trouble region within the service area; and
continuously updating the trouble forecast models with new weather data from one or more weather observation sources by applying data filters to filter out low confidence data from the new weather data, wherein the low confidence data comprises data that fails to meet data source-specific quality thresholds, transforming the new weather data to common weather features including at least one of relative humidity, station pressure, temperature, wind speed, wind direction, pressure, accumulated rain, or rain rate, calibrating training input features using statistical influence techniques and retraining the trouble forecast models using the calibrated new weather data;
determining a regional job forecast for a geographic service area based on the trouble forecast for the geographic service area according to a trouble-to-job mapping, wherein managing resources is based on the regional job forecast the service area; and
providing the regional job forecast for the geographic service area in order to initiate resource facilitate management of resources by relocating equipment and personnel within the service area according to the regional job forecast, in order to minimize an impact of the trouble forecast.
11 . The system of claim 10 , wherein the impact comprises an amount of time required to resolve weather-related damage to the resources.
12 . The system of claim 10 , wherein the impact comprises an amount of time required to resolve a weather-related interruption to service in the service area.
13 . The system of claim 10 , wherein the training the plurality of trouble forecast models is further based on trouble history records.
14 . The system of claim 10 , further comprising collecting historical weather data from one of a National Oceanic and Atmospheric Administration (NOAA) Metar station or a NOAA Mesonet station.
15 . The system of claim 10 , wherein determining clustered regions further comprises generating cluster regions based on similarity of the trouble regions in at least one of job type or distance.
16 . The system of claim 10 , wherein determining the regional job forecast for the service area based on the service area comprises determining a forecasting horizon for each trouble region and each type of weather-related damage.
17 . The system of claim 16 , wherein the at least one processor is configured to determine the regional job forecast for the service area based on the service area further comprises the at least one processor configured to compute, for each trouble region and type of weather-related damage, a trouble count forecast based on a probability distribution.
18 . The system of claim 17 , wherein determine the regional job forecast for the service area based on the service area further comprises computing, for each trouble region, trouble type and a plurality of prediction periods, a trouble count forecast based on a probability distribution.