| Title: | Closest History Flow Field Forecasting for Bivariate Time Series |
|---|---|
| Description: | The software matches the current history to the closest history in a time series to build a forecast. |
| Authors: | Patrick Fleming |
| Maintainer: | Kyle A. Caudle <[email protected]> |
| License: | GPL-3 |
| Version: | 0.1.0 |
| Built: | 2024-11-01 11:24:23 UTC |
| Source: | https://github.com/cran/CHFF |
Matches the current history with the “closest” history for a given time series. A forecast will be based on what happened after the “closest” history was observed.
CHFF(data,num,step)CHFF(data,num,step)
data |
Time series data |
num |
Number of forecasts produced |
step |
Step size (or lag) in past slopes used in the history structure |
Prints the (x,y) forecast values and provides a plot
Patrick Fleming
Caudle, KA, Fleming, PS, Frey, MR and Brubaker, N. "Next Generation of Flow Field Forecasting", Proceedings of the Joint Statistical Meetings of the American Statistical Association, Seattle, WA., 8 August-13 August (2015).
Frey, Michael R., and Kyle A. Caudle. "Flow field forecasting for univariate time series." Statistical Analysis and Data Mining (2013).
data(tsdata) # Load time series data int R CHFF(tsdata,10,3)data(tsdata) # Load time series data int R CHFF(tsdata,10,3)
Extracts the history space for a given time series
historyslopes(x,y,step,ave)historyslopes(x,y,step,ave)
x |
x values for the time series |
y |
y values for the time series |
step |
The lags in past slopes used in the given history |
ave |
Then number of time step the slope are averaged over. We recomend 1 or step which is the lags in the slopes. |
Returns the history space in matrix form 16X(datalength-7*step), with the most recent history at the bottom.
Patrick Fleming
Caudle, KA, Fleming, PS, Frey, MR and Brubaker, N. "Next Generation of Flow Field Forecasting", Proceedings of the Joint Statistical Meetings of the American Statistical Association, Seattle, WA., 8 August-13 August (2015).
Frey, Michael R., and Kyle A. Caudle. "Flow field forecasting for univariate time series." Statistical Analysis and Data Mining (2013).
data(tsdata) # Load time series data int R CHFF(tsdata,10,3)data(tsdata) # Load time series data int R CHFF(tsdata,10,3)
For each history we calculate the standard distance score between the current history and all histories
standarddistance(char,History,hlength)standarddistance(char,History,hlength)
char |
The number iof characteristics to consider when searching for the ”closest” history. 16 is the complete set of possible characters, 14 leaves off the x and y posistions on uses 7 x slopes and 7 y slopes only. |
History |
The history space |
hlength |
The length of the history space. |
Returns the winning Score, the structures used in the winning score, and the winning history
Patrick Fleming
Caudle, KA, Fleming, PS, Frey, MR and Brubaker, N. "Next Generation of Flow Field Forecasting", Proceedings of the Joint Statistical Meetings of the American Statistical Association, Seattle, WA., 8 August-13 August (2015).
Frey, Michael R., and Kyle A. Caudle. "Flow field forecasting for univariate time series." Statistical Analysis and Data Mining (2013).
data(tsdata) # Load time series data int R CHFF(tsdata,10,3)data(tsdata) # Load time series data int R CHFF(tsdata,10,3)
A simulated time series data model generates trajectories in a two-dimensional space. Generated trajectories are composed of 20-observation cycles, each cycle with four quarter-ellipse segments of five observations.
data(tsdata)data(tsdata)
Provides user data to run as an example
Patrick Fleming
Caudle, KA, Fleming, PS, Frey, MR and Brubaker, N. "Next Generation of Flow Field Forecasting", Proceedings of the Joint Statistical Meetings of the American Statistical Association, Seattle, WA., 8 August-13 August (2015).
Frey, Michael R., and Kyle A. Caudle. "Flow field forecasting for univariate time series." Statistical Analysis and Data Mining (2013).
data(tsdata) # Load time series data int Rdata(tsdata) # Load time series data int R