| 主权项 |
1. A method for forecasting busy hour traffic for a network, comprising:
obtaining, by a processor, actual traffic data for a seed time frame for each of a plurality of sectors in a given market and marketing projections for a plurality of forecast time frames beyond the seed time frame for the given market, wherein the actual traffic data for the seed time frame comprises historical busy hour traffic data and wherein the marketing projections for the plurality of forecast time frames beyond the seed time frame comprise marketing projections for busy hour traffic data for the plurality of forecast time frames; determining, by the processor, a market traffic actual measure for the seed time frame and for the given market from the actual traffic data for the seed time frame, wherein the market traffic actual measure is determined for the given market by summing the actual traffic data for the seed time frame across all sectors in the given market during the seed time frame; obtaining, by the processor, for the given market, a first set of market traffic estimates from a plurality of previous runs for the seed time frame, and a second set of market traffic estimates from the previous runs for the forecast time frames beyond the seed time frame; and updating, by the processor, the second set of market traffic estimates for the forecast time frames beyond the seed time frame to obtain updated market traffic estimates, wherein the updated market traffic estimates are used for forecasting the busy hour traffic data for each of the forecast time frames beyond the seed time frame for each of the sectors in the given market, wherein each of the updated market traffic estimates is determined using a first component and a second component, wherein the first component is derived by scaling the market traffic actual measure for the given market, wherein the scaling the market traffic actual measure for the given market is performed in accordance with a ratio of a marketing projection for one of the forecast time frames beyond the seed time frame to a marketing projection for the seed time frame, and wherein the second component is derived from the first set of market traffic estimates from the previous runs for the seed time frame, wherein the second component is based upon a ratio of the market traffic actual measure for the seed time frame to a market traffic estimate from the first set of market traffic estimates from the plurality of previous runs for the seed time frame; wherein the second component is derived by applying a modulator, wherein the modulator is an operator that either slows down or speeds up a scaling from one of the previous runs; wherein the modulator is a function defined as:
Modulated MTEiL=ƒ(MTEiL,β,αi), wherein a value of the function is obtained by performing:
Set β=Minimum(β,1), (1)If (αi≧1) then Modulated MTEiL={(αi−1)β+1}×MTEiL, (2)Else Modulated MTEiL={αiβ}×MTEiL,wherein an input MTEiL is the market traffic estimate for the same time frame from a previous run immediately prior to the current run, β is determined as a ratio of the actual busy hour data for a time period that has already been observed and a corresponding estimate from the previous run, and αi determined as a ratio of a current marketing projection for a specific time frame to the previous marketing projection for the same specific time frame from the previous run. |
