Identify systematic biases and improve the demand allocation model
Right now the parameters conform to the normalized population mapping matrix. This shows good prediction accuracy at the national level but there is state-level-bias which needs to be addressed.
Develop functions to create weights which adjust the demand allocation parameters: Presently, the parameter values are defined by the average population within the intersection of a demand area and census tract. The actual demand could possibly be skewed by the difference in types of energy uses e.g. residential, commercial, transportation energy.
Develop a testing methodology to check the accuracy of the new weights: The new weights will be created to handle the systematic biases and then the modified values will be validated using a training and testing methodology.
cmgosnell
commented
3 years ago
Identify systematic biases and improve the demand allocation model
Right now the parameters conform to the normalized population mapping matrix. This shows good prediction accuracy at the national level but there is state-level-bias which needs to be addressed.
Develop functions to create weights which adjust the demand allocation parameters: Presently, the parameter values are defined by the average population within the intersection of a demand area and census tract. The actual demand could possibly be skewed by the difference in types of energy uses e.g. residential, commercial, transportation energy.
Develop a testing methodology to check the accuracy of the new weights: The new weights will be created to handle the systematic biases and then the modified values will be validated using a training and testing methodology.