Computer-implemented method of predicting energy use for a route

A computer program comprising instructions which, when the computer program is executed by a computer, cause the computer to carry out a computer-implemented method of predicting energy use for a route comprising:

• inputting map data of roads included in K trips in a geographical area, and inputting predictors of rate of energy use along the roads;
• inputting energy consumption data of the K trips, the energy consumption data indicating total energy use T between a start point A and an end point B of each of the K trips;
• dividing each of the roads in the map data for all the K trips into segments of length measure λi;
• grouping the segments from the K trips into a number N of clusters, the clusters being defined in accordance with ranges of at least one of the predictors and each cluster being defined as having a weight Wj which is to be determined;
• using an algorithm to build a model predicting the weight Wj based on solving a system of equations, one per trip, each equation equating a known total energy use T of a trip with a sum of a known length measure of each segment in the trip multiplied by a weight for the cluster into which a segment was grouped;
• for each segment, assigning the predicted weight Wj applied to the cluster in which the segment was grouped as a predicted rate of energy use Yi;
• and
• storing a segment ID with an indication of the predicted rate of energy use Yi to allow prediction of energy use for a route in the geographical area incorporating one or more of the segments.

A computer-implemented method of estimating individual components that contribute to an aggregated statistic, comprising:

• inputting data in K trials, and inputting predictors of overall rate of contribution to the aggregated statistic;
• inputting contribution data of the K trials, the contribution data indicating total contribution T between a start and an end of each of the K trials;
• dividing the each of the K trials into time or space segments of length measure λi;
• grouping the segments from the K trials into a number N of clusters, the clusters being defined in accordance with ranges of at least one of the predictors, each cluster being defined as having a weight Wj which is to be determined;
• using an algorithm to build a model predicting the weight Wj based on solving a system of equations, one per trial, each equation equating a known total contribution T of a trial with a sum of a known length measure of each segment in the trial multiplied by the weight for a cluster into which the segment was grouped;
• for each segment, assigning the predicted weight applied to the cluster in which the segment was grouped as a predicted rate of contribution Yi;
• and
• storing a segment ID with an indication of the predicted rate of contribution Yi to allow prediction of contribution for an aggregated statistic incorporating one or more of the segments.